broad-scale resource selection and food habits of a
TRANSCRIPT
BROAD-SCALE RESOURCE SELECTION AND FOOD HABITS OF A RECENTLY
REINTRODUCED ELK POPULATION IN MISSOURI
___________________________________________
A Thesis
Presented to
The Faculty of the Graduate School
At the University of Missouri-Columbia
_____________________________________________
In Partial Fulfillment
of the Requirements for the Degree
Master of Science
____________________________________________
By
TRENTON N. SMITH
Dr. Joshua J. Millspaugh, Thesis Supervisor
MAY 2015
The undersigned, appointed by the dean of the Graduate School, have examined the
thesis entitled
BROAD-SCALE RESOURCE SELECTION AND FOOD HABITS OF A RECENTLY
REINTRODUCED ELK POPULATION IN MISSOURI
Presented by Trenton Smith
A candidate for the degree of
Master of Science in Fisheries
And hereby certify that, in their opinion, it is worthy of acceptance.
__________________________
Professor Joshua Millspaugh
__________________________
Professor Matthew Gompper
__________________________
Professor Hong He
_________________________
Dr. Lonnie Hansen
ii
ACKNOWLEDGEMENTS
First of all, I would like to thank my God. He called me out to Missouri and has
brought to completion that which He began in me. Without the Lord, none of this would
have been possible. It was only accomplished by His grace.
I would like to thank my advisor, Dr. Joshua Millspaugh, for giving me the
opportunity to pursue a Master’s degree in Wildlife Sciences. His wisdom, guidance,
experience, and care made my Master’s research and education a success. I would also
like to thank Dr. Lonnie Hansen, who along with Dr. Millspaugh, made this elk
restoration and research possible. I am especially grateful for Dr. Hansen walking
alongside me and providing support in this process. I appreciate the commitment of my
committee members, Drs. Hansen, Gompper, and He, to strengthen my research and
provide guidance in the development of this thesis.
I would also like to thank the current and previous members of the Millspaugh
lab. Dr. Barbara Keller and Amy Bleisch, laid the groundwork for this research by
developing and implementing sound data collection methods before I arrived in Missouri.
Dr. Keller also taught me how to perform microhistological analysis and supervised data
collection throughout the project. Dr. Christopher Rota and Tom Bonnot helped me
organize my data and provided code and guidance for my resource selection analysis. I
could not have completed this analysis without their superior knowledge and experience.
Aleshia Fremgen, Jaymi Lebrun, Jesse Kolar, Jenny Cunningham, Hemanta Kafley, Rich
Stanton, Brian Hiddon, and Sean O’Daniels, provided encouragement and necessary
iii
distractions throughout my graduate career. I am very grateful for all of their support in
the field and in the office.
We had an amazing crew of technicians that did a fabulous job of data collection:
Seth Snow, Tanya Wolf, Tim Schrautemeier, Cari Sebright, Jacob Dillon, Jana Ashling,
Dakota Neel, Stephanie Raiman, Jennifer Foggia, Jake Behrens, Matthew Thomas, Chloe
Wright, Derek Payne, Damon Haan, Nick Oakley, Kourtney Stonehouse, Josh Leonard,
Deanna Russell, Sam Garrett McKee, Brock Catalono, Denver Long, Molly Elderbrook,
Ciera Rhodes, Ian Evans, and Michael Ottenlips. I truly appreciate their dedication to
this project. I would like to especially thank our crew leader, Nick Oakley for working
hard to keep everyone organized.
Many employees of the Missouri Department of Conservation participated in this
project. I would like to personally thank the staff of Peck Ranch Conservation Area
including Preston Mabry, Marshall Price, and Patti Vessels that were always there and
willing to help when we needed them. I would also like to thank Ryan Houf for sharing
his knowledge of the area and current habitat management. Susan Farrington and
Elizabeth Olson shared their vast knowledge of Ozark flora to help with plant
identification, phenology, and the development of forage classes. I would like to thank
Matthew Olson for providing hard mast data. Lastly, I would like to thank Tim Bixler
and Joel Sartwell for developing GIS layers and software that made my job much easier.
I would also like to thank the Rocky Mountain Elk Foundation, Missouri Department of
Conservation, and the U.S. Wildlife Restoration Program for funding this research.
The Washington State University Wildlife Habitat and Nutrition Lab processed all
of my diet samples. I would like to thank Dr. Bruce Davitt for his answers to my many
iv
questions and guidance concerning digestibility analysis and the development of
correction factors. I could not have completed my digestibility analysis without the help
of Ann Kenny in the Animal Sciences Department. She harvested cattle rumen samples
and led me in the process of grinding and digesting my plant samples.
Lastly, I would like to thank my wife and family! I am so grateful for the
unconditional love and support of my parents in this process. Thank you for the amazing
opportunities that you opened up to me both in my undergraduate and graduate career.
Without them I would have never made it this far! I would like to thank my wife, Katie,
for always being here with me, for putting up with me being gone for the first summer of
our marriage, and most of all for your love and encouragement. I could not ask for a
better teammate for life!
v
TABLE OF CONTENTS
ACKNOWLEDGEMENTS ................................................................................................ ii
LIST OF TABLES AND FIGURES................................................................................. vii
ABSTRACT ....................................................................................................................... ix
Chapter I: Broad-scale Resource Selection of a Recently Reintroduced Elk Population in
Missouri .............................................................................................................................. 1
ABSTRACT .................................................................................................................... 1
INTRODUCTION .......................................................................................................... 2
STUDY AREA ............................................................................................................... 6
METHODS ..................................................................................................................... 9
Reintroduction and Field Methods.............................................................................. 9
Resource Covariates.................................................................................................. 10
Resource Selection Analysis ..................................................................................... 13
RESULTS ..................................................................................................................... 15
DISCUSSION ............................................................................................................... 17
MANAGEMENT IMPLICATIONS ............................................................................ 22
LITERATURE CITED ................................................................................................. 23
TABLES ....................................................................................................................... 32
FIGURES ...................................................................................................................... 34
vi
Chapter II: Food Habits and Diet Selection of a Recently Reintroduced Elk Population in
Missouri ............................................................................................................................ 47
ABSTRACT .................................................................................................................. 47
INTRODUCTION ........................................................................................................ 48
STUDY AREA ............................................................................................................. 51
METHODS ................................................................................................................... 53
Reintroduction........................................................................................................... 53
Fecal Sample Collection and Analysis ..................................................................... 54
Correction Factors ..................................................................................................... 55
Forage Availability ................................................................................................... 58
Diet Selection Analysis ............................................................................................. 62
RESULTS ..................................................................................................................... 64
DISCUSSION ............................................................................................................... 67
MANAGEMENT IMPLICATIONS ............................................................................ 71
LITERATURE CITED ................................................................................................. 73
TABLES ....................................................................................................................... 79
FIGURES ...................................................................................................................... 83
APPENDIX I. SEASONAL DIET COMPOSITION AND AVAILABILITY OF PLANT
SPECIES OBSERVED IN THE MISSOURI OZARKS. ................................................. 87
vii
LIST OF TABLES AND FIGURES
Chapter I: Broad-scale Resource Selection
Table
1. Covariates used to model broad-scale elk resource selection. ...................................... 32
2. Definitions of the 6 land cover types used in broad-scale elk resource selection
analysis. ............................................................................................................................. 33
Figure
1. Missouri elk restoration zone map depicting land ownership. ..................................... 34
2. Map of Peck Ranch Conservation Area. ....................................................................... 35
3. Histogram demonstrating the distribution of GPS locations of elk by time of day. ..... 36
4. Radius of available habitat method. .............................................................................. 37
5. Estimated relative probability (± 95% credible intervals) of an elk using a sample unit
based on land cover class. ................................................................................................. 38
6. Estimated relative probability (± 95% credible intervals) of an elk using a sample unit
based on canopy cover, interspersion juxtaposition index, slope, and years since
prescribed fire. .................................................................................................................. 39
7. Estimated relative probability (± 95% credible intervals) of an elk using a sample unit
based on distance to nearest two-track road and distance to nearest paved road. ............ 40
8. Estimated relative probability (± 95% credible intervals) of an elk using a sample unit
based on distance to nearest public gravel road, road density, aspect, and distance to edge.
........................................................................................................................................... 41
9. Estimated relative probability (± 95% credible intervals) of an elk using a sample unit
based on land cover class during summer, fall, winter, and spring, respectively. ............ 42
10. Estimated relative probability (± 95% credible intervals) of an elk using a sample
unit during each season as a function of distance to the nearest wooded edge. ............... 43
11. Estimated relative probability (± 95% credible intervals) of an elk using a sample
unit during each season as a function of road density. ..................................................... 44
12. Estimated relative probability (± 95% credible intervals) of an elk using a sample
unit during each season as a function of the interspersion juxtaposition index (IJI). ....... 45
viii
13. Estimated relative probability (± 95% credible intervals) of a female or male elk
using a sample unit as a function of distance to two-track road and road density. ........... 46
Chapter II: Food Habits
Table
1. Seasonal diet composition (%) before and after digestion trials and correction factors
(CF) for elk in Missouri .................................................................................................... 79
2. Seasonal diet selection ranks from summer 2011 - winter 2012 for elk in Missouri. .. 80
3. Seasonal diet composition (%) by forage class for elk in Missouri, 2011-2012. ......... 81
4. Seasonal diet composition (%) of cultivated forage plants for elk in Missouri summer
2011-winter 2012. ............................................................................................................. 82
Figure
1. Missouri elk restoration zone map depicting land ownership. ..................................... 83
2. Map of Peck Ranch Conservation Area. ....................................................................... 84
3. Representation of placement of quadrats along transects at vegetation sampling points.
........................................................................................................................................... 85
ix
ABSTRACT
Since being extirpated from eastern North America, elk (Cervus elaphus) have
been successfully restored in 8 eastern states and 1 Canadian province and have recently
been reintroduced in Missouri and Virginia. Benefits of elk reintroduction include
restoring ecosystem function, increasing recreation opportunities (wildlife-viewing and
hunting), and potentially boosting local economies. Thus, there continues to be strong
interest in elk restoration activities in other states.
Understanding elk habitat needs is crucial for habitat management of restored
populations and for designating appropriate reintroduction sites for new populations.
Habitat needs of western populations of elk have been well defined. However, relatively
little is known about the habitat needs of elk in eastern North America. Therefore, our
objectives were to determine broad-scale resource selection and food habits of the
recently reintroduced elk population in Missouri. To achieve these objectives, we placed
GPS collars on all adult animals prior to their release which collected locations every 4-5
hours.
To determine elk resource selection, we first defined resource attributes using GIS
layers of various landscape features. We recorded resource use using elk locations from
GPS collars and defined availability for each location according to the movement
potential of elk. We modeled resource selection using a hierarchical Bayesian discrete
choice model including interactive effects of season (summer, fall, winter, and spring)
and sex. We modeled the relative probability an elk would choose a resource from
among a set of alternatives as a function of 9 resource attributes including vegetation
type, distance to road, road density, prescribed fire, slope, aspect, canopy cover, edge,
x
and interspersion. The positive effect of forage openings (fields cultivated to provide
forage for wildlife) on elk resource selection was overwhelmingly greater than all other
landscape features across all seasons. Elk also selected cool-season grasslands (pastures)
over other vegetation types, except forage openings. The relative probability of use by
elk increased as interspersion of vegetation types increased and the percentage of canopy
cover decreased. Elk avoided paved roads, but selected areas near two-track roads.
These patterns in elk resource selection were consistent across seasons. However, elk
exhibited selection for glades, cool-season grasslands, and more heterogeneous
landscapes mostly during winter and spring. The availability of open lands (areas with
essentially 0% canopy cover including glades, pastures, and forage openings) is a
critically important resource for elk in forest dominated landscapes in the eastern U.S.
Managers of elk populations in similar ecosystems should ensure the availability of open
lands is sufficient which might be met through maintenance of forage openings and
restoration of natural open lands. Even-aged forest management that incorporates small
clear-cuts may also improve elk habitat conditions.
We determined the seasonal diet selection of elk in Missouri by comparing diet
composition (use) with forage availability of 12 forage classes. We measured diet
composition through microhistological analysis of feces. We collected feces from June
2011 through February 2013 by randomly selecting an individual elk 3 times per week
and searching for fresh fecal samples near the most recent location from its GPS collar.
We determined forage availability through vegetation sampling at 201 random points
stratified by vegetation type. Elk selected grains and cool-season grasses over all other
forage classes across all seasons and years except during summer 2012 and fall 2012.
xi
Legumes were the most highly consumed forage class for all seasons and years. Native
forbs were ranked as highly selected during the 2012 drought (summer and fall), but were
not selected by elk during 2011. Approximately half (44.6%) of the annual elk diet was
composed of plant species cultivated in wildlife forage openings. Clover (Trifolium
spp.), alfalfa (Medicago sativa), and common lespedeza (Kummerowia spp.) were the
most highly consumed species by elk. In forested dominated landscapes, we recommend
the creation of forage openings composed of cool-season grasses and legumes to increase
forage availability for elk. Management for natural open lands (e.g. glades) composed of
native forbs and grasses may provide an alternative food source when food is scarce.
1
Chapter I: Broad-scale Resource Selection of a Recently Reintroduced Elk
Population in Missouri
ABSTRACT
Elk (Cervus elaphus) have been successfully reintroduced in 8 U.S. states and 1
Canadian province in eastern North America and have recently been reintroduced in
Missouri and Virginia. Resource selection has been well defined for western populations
of elk. However, there is limited research concerning resource selection of eastern elk
populations. Our objective was to determine broad-scale resource selection of recently
reintroduced elk in the Missouri Ozarks. We placed GPS collars on all adult animals
prior to their release which collected locations every 4-5 hours. We defined 9 resource
attributes using GIS layers of various landscape features including vegetation type,
distance to road, road density, prescribed fire, slope, aspect, canopy cover, edge, and
interspersion. We modeled resource selection using a hierarchical Bayesian discrete
choice model including interactive effects of season (summer, fall, winter, and spring)
and sex. We modeled the relative probability an elk would choose a resource from
among a set of alternatives as a function of its resource attributes. The positive effect of
forage openings (fields cultivated to provide forage for wildlife) on elk resource selection
was overwhelmingly greater than all other landscape features across all seasons. Elk also
selected cool-season grasslands (pastures) over other vegetation types, except forage
openings. The relative probability of use by elk increased as interspersion of vegetation
types increased and the percentage of canopy cover decreased. Elk avoided paved roads,
but selected areas near two-track roads. These patterns in elk resource selection were
2
consistent across seasons. However, elk exhibited selection for glades, cool-season
grasslands, and more heterogeneous landscapes mostly during winter and spring. The
availability of open lands (areas with essentially 0% canopy cover including glades,
pastures, and forage openings) is a critically important resource for elk in forest
dominated landscapes. Managers of elk populations in similar ecosystems should ensure
the availability of open lands is sufficient which might be met through maintenance of
forage openings and restoration of natural open lands. Even-aged forest management that
incorporates small clear-cuts may also improve elk habitat conditions.
INTRODUCTION
Many wildlife species were extirpated from all or parts their native range due to
overharvest and habitat loss. As native species began to disappear, conservationists
sought solutions for restoring these species. One such solution was translocation of
individuals from areas where the species still existed to areas where the species had been
extirpated (i.e. reintroduction; Griffith et al. 1989). Some of the success stories of
wildlife reintroductions in North America include wild turkeys (Meleagris gallapavo),
bighorn sheep (Ovis canadensis), gray wolves (Canis lupus), and elk (Cervus elaphus)
(Griffith et al. 1989, Wolf et al. 1996, Ripple and Beschta 2004). These reintroductions
are important because, by restoring a species to its native habitat, the reintroductions may
promote a more intact and functioning ecosystem (Ripple and Beschta 2004, Peled 2010).
Since being extirpated from eastern North America, elk have been successfully
restored in 8 eastern states and 1 Canadian province (Arkansas, Kentucky, Michigan,
Minnesota, North Carolina, Pennsylvania, Tennessee, Wisconsin, and Ontario; Popp et al.
3
2014), with 2 recent reintroductions in Missouri and Virginia. Although ecosystem
restoration may be a seemingly intangible benefit, elk restoration also offers opportunities
for hunting and wildlife viewing that otherwise would only be available in the western
United States (MDC 2010). Moreover, this increase in recreational interest might benefit
local economies by bringing more visitors to the area (RMEF and Southern & Eastern
Kentucky Tourism Development Association 2007). Alabama, West Virginia, and
Maryland have recently expressed interest in reintroducing elk. Wisconsin is currently
augmenting their populations through translocations and Minnesota is considering such.
Thus, there continues to be strong interest in elk restoration activities.
Managers consider several factors when determining where to reintroduce elk
including habitat availability, land ownership, and potential for human conflict.
However, there is limited research concerning habitat needs for elk in the eastern United
States. Thus, habitat availability for eastern reintroductions has been estimated mostly
based on our understanding of suitable habitat for elk in the western United States where
ecological conditions are much different (Van Deelen et al. 1997, McClafferty 2000).
Lack of information regarding elk habitat in eastern populations also limits informed
management of areas where elk have already been restored. Therefore, further research
of elk habitat needs in the eastern United States is essential to select appropriate
reintroduction sites and to implement habitat management once elk are released.
Understanding how an animal selects available resources (i.e., resource selection)
is critical to making decisions concerning management of a species and its habitat.
Furthermore, determining resource selection is important for establishing new
populations, because resource selection identifies landscape attributes that are important
4
to the species which will ultimately help determine the distribution and success of the
new population. Several studies examined reproduction, mortality, survival, site fidelity,
and food habits of eastern populations of elk, particularly in Kentucky (Larkin et al.
2003, Larkin et al. 2004, Schneider et al. 2006, Lupardus et al. 2011). However, research
examining broad-scale resource selection of eastern populations is limited. Anderson et
al. (2005) determined resource selection of elk in Wisconsin during summer, but not for
the remainder of the year. Resource selection of elk has also been studied in Kansas
(Conard 2009), Oklahoma (Walter 2006), and Ontario (Popp et al. 2013). However, the
vegetation communities of these areas differ greatly from the deciduous forests of many
eastern populations. Research shortly after elk were released in Kentucky provided some
insight about the habitat requirements of eastern populations, but focused on site fidelity,
rather than resource selection (Larkin et al. 2004). Telesco et al. (2007) examined elk
habitat use in Arkansas, but the main purpose of this study was to determine other
potential reintroduction sites and it did not explicitly examine resource selection. Two
resource selection studies were conducted on the reintroduced elk population in Great
Smoky Mountains National Park, North Carolina (Murrow 2007, Hillard 2013).
However, only Murrow (2007) determined annual elk resource selection. Therefore, a
review of the literature demonstrates that research regarding broad-scale elk resource
selection for restored populations is lacking for eastern deciduous forests.
In Missouri, there is a great need to understand resource selection for the newly
restored elk population, so state managers can improve habitat conditions necessary to
sustain the herd and minimize human-elk conflicts. The Missouri Department of
Conservation (MDC) has focused on the management of woodlands (<80% canopy
5
closure), glades, and wildlife forage openings (fields cultivated to provide forage for
wildlife) within the elk restoration area and considers these areas as suitable habitat for
many wildlife species including elk (MDC 2010). In their reintroduction plan, MDC
proposed to maintain at least 10% of the restoration zone in woodlands, glades, and open
lands (areas with essentially 0% canopy cover including glades, pastures, and forage
openings), but it is unknown whether elk will select these vegetation types and what other
factors might affect their distribution. It is crucial to assess elk use of these areas to
determine whether MDC should continue to pursue this goal. This information is
important not only to the management of the population in Missouri, but also has
implications for other eastern populations of elk in similar habitats.
Elk are known to use a wide variety of landscape types (Skolvin 2002). Despite
the diversity in habitat use, there are consistent trends in habitat selection studies for elk.
Elk often select open lands and areas with a high amount of interspersion of habitat types
(i.e., edge; Irwin and Peek 1983, Skolvin 2002, Anderson et al. 2005, Murrow 2007,
Baasch et al. 2010). Highest site fidelity for reintroduced elk in Kentucky occurred in
areas with the greatest edge density (5 km/km2) (Larkin et al. 2004). Baasch et al. (2010)
observed that the most important factor affecting elk resource selection in Nebraska was
distance to ponderosa pine (Pinus ponderosa) edge. Many studies also cite forage
availability and quality as an important factor in elk resource selection (Unsworth et al.
1998, Anderson et al. 2005, Long et al. 2009). Elk are also known to select the cover of
closed canopy forests for thermoregulation (Millspaugh et al. 1998) and to avoid
predators and human disturbance (Unsworth et al. 1998). Therefore, a combination of
6
habitat features related to food and cover are typically overriding features in resource
selection of elk.
Our objective was to assess broad-scale resource selection of the recently
reintroduced elk population in Missouri. We hypothesized that thermal cover, forage
availability, and human disturbance would influence elk resource selection. We built a
population-level resource selection model with multiple resource parameters to test these
hypotheses. We determined elk resource selection accounting for the potential effects of
season and sex.
STUDY AREA
In 2010, MDC designated an 896 square-kilometer elk restoration zone in
Shannon, Reynolds, and Carter counties in southeast Missouri (Figure 1) (MDC 2010).
This area was selected based on potential availability of elk habitat, a high percentage of
public land ownership, few row crops, and limited public roads (MDC 2010). The elk
restoration zone is located in the Missouri Ozarks and is composed mostly of oak-hickory
and oak-pine forests with the landscape being 93% forested, 5% open, and 0.1% cropland
(Xu et al. 1997, MDC 2010). Open lands are made up of glades dominated by warm-
season grasses and forbs, forage openings planted with cool-season and warm-season
grasses and legumes, and pastures (MDC 2010, Tousignant 2011). The majority of the
land (79%) is open to the public with 49% in public ownership and managed by MDC,
the United States Forest Service, and the National Park Service and 30% owned and
managed by The Nature Conservancy and the L-A-D Foundation (a local private
organization; MDC 2010).
7
The elk restoration zone is located in the Ozark Highlands region of Missouri
which is made up of a mosaic of ridges and valleys with many caves and springs (Meinert
et al. 1997). Soils in this region range from deep loamy soils to shallow rocky soils
generally with dolomite or sandstone substrates with some igneous rhyolite substrates on
higher hills (Meinert et al. 1997). Ongoing forest management activities include small
clearcuts (<12 ha), selection cuts, and thinning. Prescribed fire is utilized throughout the
Ozarks on large burn units (average 300 ha) on a relatively regular rotation of 3 years.
Extensive habitat management has led to a diverse array of forest types including closed
canopy forests, open canopy woodlands with an understory of forbs and grasses, and
regenerating forests with dense shrubs and saplings. The Current River bisects the elk
restoration zone with many permanent and intermittent streams flowing into it.
The mean annual temperatures for our study area for 2011, 2012, and 2013 were
12.3, 13.7, and 14.9 ̊ C, respectively, with maximum temperatures occurring in July (30,
35, and 37 ̊C) and minimum temperatures occurring in January (-4, -5, and -7 ̊ C) (Prism
Climate Group, www.prism.oregonstate.edu, accessed 23 Sept. 2014). Most
precipitation in this area is in the form of rainfall. A severe drought occurred throughout
the Midwestern United States including our study area during the summer of 2012.
Limited rainfall was most evident during June. June precipitation in our study area was
slightly less than the 30-year (1981-2010) average (97mm) during 2011 (71mm) and
slightly greater than average in 2013 (121mm; Prism Climate Group 2014). June
precipitation in 2012 was less than half of the 30-year average (47mm; 48% of average
precipitation; Prism Climate Group 2014).
8
Peck Ranch Conservation Area near Van Buren, Missouri was the primary
location for the elk reintroductions, and where most of the released elk were located
(Figure 2). This is a 9,327 hectare parcel managed by MDC (MDC 2013). A large
portion of this conservation area (4,283 hectares) is maintained as a wildlife refuge where
hunting is only allowed during three managed white-tailed deer (Odocoileus virginianus)
hunts annually. Within Peck Ranch, there is a relatively extensive network of gravel
roads. However, since the elk were first released in 2011, MDC designated some of these
roads as a public elk viewing route with the remaining roads closed to public vehicles
(Figure 2). MDC also closes the entire wildlife refuge to all public vehicle traffic from
April 1 through July 1 to prevent disturbance of the elk during the calving season.
Elevation on most ridges in Peck Ranch ranges from 275 to 300 meters with a
maximum elevation of 411 meters on Stegall Mountain (MDC 2013). This area contains
2 main streams, Mill and Rogers Creeks and over 100 small ponds. Like the rest of the
restoration zone, Peck Ranch is composed primarily of oak-dominated forest and
woodland (91%) with 9% open lands (MDC 2013). MDC has made it a management
goal to increase, restore, and maintain woodlands, glades, and forage openings for
wildlife including elk (MDC 2010). MDC’s management effort is particularly evident on
Peck Ranch Conservation Area. Prescribed fire has been utilized extensively on Peck
Ranch with 15% of the area being burned on average each year resulting in the
restoration of many woodlands and glades. Restoration of glades and maintenance of
over 100 forage openings, ranging in size from 0.05ha to 8ha, has resulted in a much
greater proportion of open lands on Peck Ranch (9%) compared to that of the elk
restoration zone overall (5%) (MDC 2010, MDC 2013).
9
METHODS
Reintroduction and Field Methods
The initial reintroduction of elk in Missouri occurred in the summer of 2011. All
elk were captured and translocated from eastern Kentucky to Peck Ranch Conservation
Area in Missouri where they were held for 4 weeks in 4 pens, each approximately 1 acre
in size, to acclimate to the new environment. On June 1, 2011, 33 adult elk (19 females,
14 males) were released onto Peck Ranch Conservation Area from the pens. The
following summer, elk were released at 2 sites. Twenty female adult elk were released at
the Peck Ranch site on June 19, 2012. On June 23, 2012, 14 adult elk (6 females and 8
males) were released on Chilton Creek Preserve owned by The Nature Conservancy
approximately 8 km northeast of the Peck Ranch site in an attempt to expand the range of
the elk population (Figure 1). On June 7, 2013, 39 more adult elk (36 females and 3
males) were released at the Peck Ranch site.
Before release, all elk ≥1 year old were fitted with a GPS collar (RASSL custom
3D cell collar, North Star Science and Technology, LLC, King George, VA, or G2110E
Iridium/GPS series model, Advanced Telemetry Systems, Insanti, MN), an ear tag (Salt
Lake Stamp Company, Salt Lake City, Utah), and a PIT tag (LifeChip Permanent
Identification, Destron Fearing, South St. Paul, Minnesota; preloaded sterile GPT12,
Biomark, Boise, Idaho). All GPS collars were set to collect and store one internal
location every 2.5 hours and upload a location to the internet every 5 hours, except 2
collars which were set to record internal locations every 2 hours and upload a location
every 4 hours. We chose this location schedule to ensure that elk locations were
10
distributed throughout all hours of the day (Figure 3). Each collar had a blow-off device
that was set to detach from the animal approximately 2 years after the animal was
released so that the collar could be recovered and the internal locations downloaded. The
collars have a VHF component that was used to locate the elk when the GPS unit failed
and to assist in retrieving dead animals. All collars were equipped with a mortality
sensor set to detect when the animal has not moved for more than 4 hours. All animal
activities were approved by the University of Missouri Animal Care and Use Committee
(Protocol 6909).
Resource Covariates
For this study, we determined third-order resource selection (selection within
home range) (Johnson 1980) of elk by comparing the resource attributes at points where
elk were located with attributes at random points considered available. We measured
resource attributes from GIS layers, which we created and manipulated with ArcInfo 10
(Environment Systems Research Institute, Inc., Redlands, CA).
We defined 9 resource attributes within 30x30m cells which we will call
“resource units” (Table 1). Through cooperation with the Missouri Resource
Assessment Partnership (MoRAP) and MDC, we developed a GIS layer of land cover for
the elk restoration zone using 2010 Landsat™ imagery, aerial photography, and ground
collected GPS points. We added forage openings to this layer using data collected by
MDC. We classified land cover into 9 classes with “other” as the reference type (Table
2). We acquired a percent tree canopy cover layer from the 2011 US Forest Service
National Land Cover Database (www.mrlc.gov/nlcd11_data.php, accessed 8 Jan 2015).
11
We used the field calculator in ArcInfo to adjust the canopy cover values in all forage
openings to 0%. We also developed a GIS layer of years since prescribed fire for our
study area based on burn units and burn years provided by MDC, the National Park
Service, and The Nature Conservancy and updated it annually in the spring to reflect
recent burns. We inserted a value of 70 years for all locations outside of the given burn
units, because the last major wildfire that occurred in this area was in 1941 (Steve
Orchard, Missouri Department of Conservation, personal communication). We acquired
digital elevation models (DEM) for the restoration zone from the Missouri Spatial Data
Information Service (MSDIS, www.msdis.missouri.edu/data/dem/index.html, accessed
13 Dec 2013) and used Surface Tools in ArcInfo to create aspect (degrees) and slope (%)
layers based on the DEM’s. To quantify the potential effect of edge, we calculated
distance to nearest wooded edge. We used the Geospatial Modeling Environment (GME;
Beyer 2012) to extract the edges between wooded (forest or woodland) and open areas
from the land cover layer and then used the Euclidean distance tool in ArcInfo to
calculate distance to edge. We also calculated the interspersion and juxtaposition index
(IJI) using Fragstats (version 4; McGarigal et al. 2012). This landscape metric
determines the spatial distribution of different types of land cover patches and increases
from 0 to 100 as patch types become more evenly distributed on the landscape (Griffith et
al. 2000). IJI was calculated within a circle defined by the average radius of available
habitat across all animals (1.922km; Durner et al. 2009). We used distance to road and
road density to account for potential human disturbance. We used 3 separate covariates
for distance to nearest road, one for each road type (paved road, public gravel road, two-
track road (roads not open to the public)), to determine whether selection is affected
12
differently by road type. We calculated road density as kilometers of paved roads and
public gravel roads per square kilometer contained within a circle with a 550m radius
using the line density tool in ArcInfo. We chose 550m, because this is the distance at
which elk perceive human disturbance from roads based on available literature (Benkobi
et al. 2004, Rumble et al. 2007). We did not include two-track roads in our road density
calculations, because Benkobi et al. (2004) and Rumble et al. (2007) observed little or no
effect by roads closed to the public. We developed our road layers from TIGER 2010
road data from the United States Census Bureau, Missouri Department of Transportation
road data, and logging road data from MDC.
We measured resource covariates within resource units where elk were located
(hereafter “used locations”) and resource units considered available to an elk. We
defined availability by randomly selecting 5 points within a circle created using the
radius of available habitat method (Durner et al. 2009). This circle was centered on the
previous used location with radius = c(a + 2b), where a is the mean hourly movement
rate (m/hr), b is the standard deviation of the movement rate, and c is the number of hours
between locations (Figure 4). We calculated monthly mean hourly movement rates to
account for variable movement rates over time (Wichrowski et al. 2005). We also
calculated separate movement rates for males and females. When the used location
occurred outside of the predetermined radius we assumed the radius of availability was
the distance between the previous and current used location (Durner et al. 2009). We
used GME (Beyer 2012) to calculate the movement rates and hours between locations.
We censored locations of all animals suspected of infection with meningeal worm
(Parelaphostrongylus tenuis; n = 13), of all animals with <30 days of telemetry data
13
during a particular season/year, and locations that occurred outside the extent of GIS
layers. We additionally censored all locations that occurred < 2 hours after the previous
location because there was uncertainty in the accuracy of these locations, and that
occurred ≥ 24 hours apart because the movement potential of an elk over this amount of
time made it impossible to accurately measure availability using the radius of available
habitat (removed 0.688% of choice sets).
Resource Selection Analysis
We modeled elk resource selection using a hierarchical Bayesian discrete choice
model (Cooper and Millspaugh 1999, Thomas et al. 2006). Discrete choice theory is
based on the assumption that when an individual has a set of resources (choice set), it will
choose the resource with attributes that result in the greatest “utility” (Cooper and
Millspaugh 1999). The utility of each resource unit (j) within choice set (i) is modeled
by:
𝑈𝑖𝑗 = 𝛽1𝑥𝑖𝑗1 + ⋯ + 𝛽𝑘𝑥𝑖𝑗𝑘
where 𝛽𝑘 is the slope coefficient associated with attribute (k) and xijk is the value of
attribute (k) at resource unit (j) in choice set (i). The probability of selecting resource unit
(j) within choice set (i) is calculated as:
𝑃𝑖𝑗 =exp (𝑈𝑖𝑗)
∑ exp (𝑈𝑖𝑗)𝐽𝑗=1
,
where J represents the number of choices in the choice set.
14
We used a hierarchical “random slopes” utility model that assumed the slope
coefficient (β) of individual elk (w) was a random variable drawn from a population-level
distribution:
𝛽𝑤𝑘~𝑁(𝜇𝑘, 𝜎𝑘2)
where μk is the population-level mean of the slope parameters of the kth
variable and σ2
k is
the associated population-level variance. We modeled the utility of each resource unit as
a function of land cover type (8 dummy variables), canopy cover, years since prescribed
fire, aspect, slope, distance to edge, IJI, distance to road (3 variables), and road density
(18 total variables). We also included interactive effects between (variables) and sex and
season. We divided seasons according to the following dates: spring (March 1- May 15),
summer (May 16- September 15), fall (September 16- November 30), and winter
(December 1-February 29). We assumed the following prior distributions for hyper-
parameters of the slope coefficients and interaction terms:
𝜇~𝑛𝑜𝑟𝑚𝑎𝑙(0,10)
𝜎~𝑢𝑛𝑖𝑓𝑜𝑟𝑚(0,10)
We made all inferences based on the population-level slope coefficients and interactions
terms (µ’s) which we will call “population-level means”.
We used RStan (Version 2.6.0, www.mc-stan.org, accessed 25 Feb 2015) and
Program R (Version 3.1.2, www.R-project.org, accessed 25 Feb 2015) to fit our discrete
choice model. We created 3 Markov chains to simulate the posterior distribution of each
population level parameter with each chain consisting of 5000 iterations including a burn-
in of 1000 iterations. We used the Brooks-Gelman-Rubin diagnostic to assess
convergence of the three chains (R̂ ≤1.1; Gelman et al. 2014).
15
We assessed goodness of fit of our model using posterior predictive checks
(Gelman et al. 2014, Kéry 2010). This approach involves comparing observed data with
predictions made by the model through 2 test statistics:
𝑡𝑒𝑠𝑡𝑜𝑏𝑠 = ∑(𝑦𝑖𝑗 − 𝑝𝑖𝑗)2
𝑝𝑖𝑗𝑖𝑗
𝑡𝑒𝑠𝑡𝑝𝑟𝑒𝑑 = ∑(𝑌𝑖𝑗 − 𝑝𝑖𝑗)2
𝑝𝑖𝑗𝑖𝑗
where yij represents the observed choice and Yij represents predicted choice simulated
from the model (Yij = 1 for the used alternative and 0 for available units). We then
compared the test statistics by calculating the Bayesian p-value of Pr(testpred > testobs).
We assumed p-values between 0.10 and 0.90 indicated adequate goodness of fit.
RESULTS
We collected GPS locations from 88 elk (65 females, 23 males) from June 1, 2011
to September 15, 2014. We modeled elk resource selection based on 141,197 choice sets
composed of 6 alternatives including the used location. We achieved convergence for all
hyper-parameters in the model (R̂ ≤1.1). We observed adequate goodness-of-fit for this
model (Bayesian p-value = 0.19).
Among all variables considered in the model, forage openings had the greatest
effect on elk resource selection (µ = 2.34, 95% CI = [2.22, 2.47]). The relative
probability of use of forage openings was over 4 times greater than all other land cover
classes (Figure 5). Cool-season grasslands also had high relative probabilities of use (µ =
0.5, 95% CI = [0.38, 0.64]; Figure 5). Elk selected areas with low canopy cover (µ = -
0.31, 95% CI = [-0.36, -0.27]; Figure 6). Relative probability of use increased as the
16
interspersion juxtaposition index (IJI) increased meaning that elk selected more
heterogeneous landscapes relative to other available areas (µ = 0.38, 95% CI = [0.31,
0.46]; Figure 6). There was a slight negative association between relative probability of
use and years since prescribed fire (µ = -0.14, 95% CI = [-0.18, -0.09]; Figure 6)
indicating more recently burned areas were selected. The relative probability of use by
elk also decreased as slope increased (µ = -0.99, 95% CI = [-0.52, -0.46]; Figure 6). Elk
selected areas that were further from paved roads and areas near two-track roads (µ =
0.26, 95% CI = [0.19, 0.33], µ = -1.05, 95% CI = [-1.3, -0.79]; Figure 7), but were not
affected by public gravel roads (µ = 0.05, 95% CI = [-0.01, 0.1]; Figure 8). Elk also
selected areas with low road density, but this relationship was not strong (µ = -0.07, 95%
CI = [-0.12, -0.03]; Figure 8). Aspect and distance to edge did not substantially affect elk
resource selection (95% credible intervals of population-level means overlap 0; Figure 8).
Elk exhibited few substantial differences in resource selection among seasons.
Forage openings were selected over all vegetation types across all seasons (Figure 9).
During summer and fall, all other vegetation types were selected at relatively equal rates
(Figure 9). Elk selected areas near edge during summer, but edge was not important the
rest of the year (Figure 10). Elk also selected low road densities during summer, but road
density did affect selection during other seasons (Figure 11). During winter and spring,
both glades and cool-season grasslands were selected at greater rates than other
vegetation types except forage openings (Figure 9). Elk also selected more
heterogeneous landscapes (higher IJI) during winter and spring (Figure 12).
We observed only 2 substantial differences between male and female resource
selection. Both male and female elk selected areas near two-track roads, however males
17
showed a much stronger relationship with two-track roads (Figure 13). Females selected
areas with low road densities, while males showed no relationship with road density
(Figure 13). However, neither of these relationships with road density was strong. All
other variables included in our analysis did not differ between males and females (95%
credible intervals of interaction terms overlap 0 or males and females exhibited the same
pattern).
DISCUSSION
Elk resource selection in Missouri was mostly driven by forage availability and
open lands. Variables associated with open lands including low canopy cover, forage
openings, and cool-season grasslands (pastures) had the greatest effect on annual elk
resource selection. Selection of open lands by elk has been documented in both eastern
and western populations (Irwin and Peek 1983, Skolvin et al. 2002, Anderson et al. 2005,
Murrow 2007). Elk will select areas where they can acquire the most forage per unit
time, which in many cases is in open lands (Cook 2002, Anderson et al. 2005). However,
foraging in open lands also elevates vulnerability to predation and hunting mortality for
elk. Therefore, elk must balance maximizing forage intake with minimizing predation
risk when selecting habitats (Wolff and Van Horn 2003). Elk in Missouri are not
constrained by either of these factors so they can forage in open lands without increased
predation risk. Thus, given these trade-offs, it is not surprising elk selected forage
openings and other open lands. The challenge for managers is the availability of open
lands in the Ozarks.
18
Open land is limited in the Missouri Ozarks with less than 5% of the elk
restoration zone composed of open land (excluding pasture on private lands). Others
have suggested that woodlands and savannas, which maintain a more open canopy than
surrounding forests, would provide habitat conditions desired by elk in eastern deciduous
forests (Lupardus et al. 2011, Telesco et al. 2007). However, although woodlands and
savannas were more readily available to elk in Missouri than open lands, they selected
open lands over these other vegetation types. Likewise, Skolvin et al. (1989) observed
that small clear-cuts (2-8ha) resulted in a substantial increase in elk use compared to
partially cut stands. We suspect that forage quantity and quality in woodlands and
savannas was not comparable to that in open lands. Forage openings within the elk
restoration zone are planted each fall with a mixture of legumes and grains, fertilized and
limed regularly, and mowed each summer to encourage fresh growth of grasses and
forbs. Pastures also experience regular maintenance, but are not usually as intensively
managed as forage openings. Glades are dominated by warm-season grasses and forbs
which provide forage for elk, but they are not managed specifically for forage production.
Therefore, forage openings likely provided the highest quality forage which is why elk
selected this vegetation type over all others. Woodlands and savannas, while providing
foraging opportunities, were not as attractive as open lands. It is possible once the elk
population grows, resulting in increased competition, that these areas could prove to be
more important, but they are currently not selected.
Clear-cuts are another means of creating open lands in forest dominated
landscapes. We did not directly quantify the effect of newly created clear-cuts on elk
resource selection, but elk selection for clear-cuts has been observed in several western
19
populations (Lyon and Jensen 1980, Irwin and Peek 1983, Skolvin et al. 1989). We
observed that the relative probability of use by elk increased as canopy cover decreased,
which may be interpreted as any opening in the canopy will increase elk use. However,
management for stands with partially open canopies (i.e. woodlands and savannas) did
not substantially increase elk use in Missouri. Removal of the majority of the overstory
canopy through small clear-cuts (< 8ha) results in greater forage production than in
partially cut stands which may increase elk use (Skolvin et al. 1989). Therefore, even-
aged forest management may be more beneficial to elk than woodland management.
The effects of human disturbance on elk distribution have been observed by many
researchers (Marcum 1975, Lyon 1979, Unsworth et al. 1998). Roads provide access for
hunters, wildlife watchers, and logging operations all of which may alter the distribution
of elk (Lyon and Christensen 2002). We observed only modest effects of roads on elk
resource selection. Our analysis indicated that elk avoided paved roads, but were not
affected by public gravel roads. Paved roads have a greater impact on the landscape than
gravel roads with larger widths, deeper road cuts, and higher vehicle traffic (Lyon and
Christensen 2002). However, traffic is also more predictable on paved roads, which may
be better tolerated by elk than inconsistent disturbances on less traveled roadways (Ward
et al. 1976). It is possible that elk in Missouri are not actively avoiding paved roads, but
are simply selecting resources that are available far from paved roads. Forage openings
had the greatest effect on elk resource selection and most forage openings in the elk
restoration zone are located in areas with few or no paved roads. Elk also selected areas
near two-track roads. Forage openings in Missouri are usually located either along or
near two-track roads. Therefore, we believe that the relationships observed between elk
20
and roads in Missouri were actually related to the location of forage openings rather than
human disturbance.
Studies of western populations of elk also indicate that cover is an important
factor affecting elk resource selection (Skolvin et al. 2002). Millspaugh et al. (1998)
observed that one of the most common characteristics of summer bed sites in South
Dakota were sites that had characteristics of thermal cover (e.g. closed canopy, lower
microsite temperature, and northern aspect). Others have observed that thermal cover is
not required by elk (McCorquodale et al. 1986, Cook et al. 1998). Elk in southern
Missouri experience warmer temperatures and higher humidity than these western
populations. Therefore, we hypothesized that thermal cover (aspect and high canopy
cover) might affect elk resource selection. However, elk selected areas with low canopy
cover across all seasons and were not affected by aspect. Elk selected areas near wooded
edge during summer, but not during other seasons. This may be evidence that elk were
selecting areas near cover to cope with high temperatures during summer. Nonetheless,
thermal cover is abundant within our study area so it does not substantially affect the
distribution of this elk population.
Elk selection for forage openings was consistent across all seasons, but they also
selected other open lands (i.e. glades and cool-season grasslands) during winter and
spring. Forage in the Missouri Ozarks is limited during winter and forage openings
receive heavy use from deer and elk during this period. Therefore, forage within forage
openings may be depleted during winter and early spring. Elk may utilize residual
grasses in pastures to supplement their diet during winter and early spring. We did not
detect this pattern in the elk winter or spring diets, because some forage species are found
21
in both forage openings and cool-season grasslands (Chapter 2). Oak browse (Quercus
spp.) was an important forage item for elk during both spring and winter, but not during
other seasons. It is possible that elk used glades to browse oak saplings during winter
when food is scarce and during spring when palatable new growth is available. We
believe that selection for more heterogeneous landscapes (higher interspersion) during
winter and spring was another indication of elk foraging in a wider variety of vegetation
types (Skolvin et al. 2002).
Our analysis detected few differences in resource selection between male and
female elk. The greatest difference was that males selected areas closer to two-track
roads than females. This is contradictory to studies of western populations that show that
male elk tend to have a stronger aversion to roads than females (Marcum 1975, Skolvin et
al. 2002). However, unlike western populations, the Missouri elk population is not
hunted. Bulls tend to separate from cows during summer and may have utilized
resources that are closer to two-track roads than the resources used by cows during this
period. No other differences were evident between bulls and cows. Bull elk in western
North America generally select steeper slopes, higher elevations, and denser cover than
female elk which has been attributed to bull elk selecting security cover to avoid hunting
mortality (Unsworth et al. 1998, Skolvin et al. 2002). Because the Missouri population is
not hunted, bull and cow elk may not differentiate in their selection for these habitat
variables. We also may not have observed many differences between the sexes, because
the collars on males ceased working more quickly than those on females so we may not
have had a large enough sample size of males to detect differences. Additionally, the
majority of males in our study were yearlings (< 2 years old), which generally do not
22
separate from cows except during the rut (Unsworth et al. 1998). Lastly, we did not
examine selection between males and females on a seasonal basis, which may have
revealed more differences.
MANAGEMENT IMPLICATIONS
Elk selection of forage openings was overwhelmingly greater than all other
landscape features. Additionally, selection for open lands and variables related to open
lands was consistent across all seasons. In densely forested landscapes occupied by
eastern populations of elk, open lands are a critically important resource. Eastern elk
managers and states and provinces considering elk reintroduction should consider the
availability of open lands when making habitat management decisions and when
selecting reintroduction sites. Many western elk managers consider habitats with a ratio
of 40% forest cover to 60% open land as optimum habitat for elk (Skolvin et al. 2002).
However, such a ratio is usually not possible at large scales in eastern deciduous forests.
Nonetheless, increasing the availability of open lands is important to provide forage for
elk and to minimize human conflicts. We recommend continuing to invest in maintaining
existing forage openings and where possible developing new forage openings within the
Missouri elk restoration zone. As the elk population expands, the need for more open
lands will only increase making it difficult to meet elk needs based on existing forage
openings alone. Therefore, managers should include other management techniques to
further increase the availability of open lands.
Our results suggest that efforts to restore and maintain glades might help provide
open land for elk and provide a secondary food source particularly during winter and
23
spring. Managers should continue to emphasize restoration of glades where they
historically occurred through removal of encroaching eastern red cedar (Juniperus
virginiana) and regular prescribed fire. We did not determine the effect of glade size in
our analysis, but we noted that elk avoided large glades in the Stegall Mountain area
which may indicate that elk selected smaller glades. This result may also be caused by
the lack of sufficient forest cover adjacent to glades in this area compared to other glades
in the restoration zone.
Although we did not assess the effect of newly created clear-cuts in our analysis,
we believe this management practice may be another tool to create open lands and
foraging areas for elk in eastern deciduous forests. We recommend considering the
implementation of even-aged forest management where timber harvests are dispersed
across the landscape creating open lands for elk. Managers should take into account that
these open lands are temporary and will only increase elk use for about 5 years (Skolvin
et al. 1989). Therefore, to effectively increase open land availability, they should either
maintain enough timber harvests to replace regenerating clear-cuts or convert some clear-
cuts to permanent open lands (i.e. forage openings). As clear-cuts are created, managers
could seed haul roads and landings with grasses and legumes to further increase forage
availability within these temporary open lands (Lyon and Christensen 2002).
LITERATURE CITED
Anderson, D. P., M. G. Turner, J. D. Forester, J. Zhu, M. S. Boyce, H. Beyer, and L.
Stowell. 2005. Scale-dependent summer resource selection by reintroduced elk in
Wisconsin, USA. Journal of Wildlife Management 69:298-310.
24
Baasch, D. M., J. W. Fischer, S. E. Hygnstrom, K. C. Vercauteren, A. J. Tyre, J. J.
Millspaugh, J. W. Merchant, and J. D. Volesky. 2010. Resource selection by elk
in an agro-forested landscape of northwestern Nebraska. Environmental
Management 46:725-737.
Benkobi, L., M. A. Rumble, G. C. Brundige, and J. J. Millspaugh. 2004. Refinement of
the Arc-Habcap model to predict habitat effectiveness for elk. Research Paper
RMRS‑RP‑51. U.S. Department of Agriculture, Forest Service, Rocky Mountain
Research Station, Fort Collins, Colorado, USA.
Beyer, H.L. 2012. Geospatial Modelling Environment (Version 0.7.2.0).
http://www.spatialecology.com/gme. Accessed 17 May 2013.
Boyce. M. S., P. R. Vernier, S. E. Nielsen, and F. K. Schmiegelow. 2002. Evaluating
resource selection functions. Ecological Modeling 157:281-300.
Brooks, S. P., and A. Gelman. 1998. General methods for monitoring convergence of
iterative simulations. Journal of Computational and Graphical Statistics 7:434-
455.
Conard, J. M. 2009. Genetic variability, habitat selection, and demography in a
reintroduced elk (Cervus elaphus) population. Dissertation, Kansas State
University, Manhattan, USA.
Cook, J. G. 2002. Nutrition and food. Pages 258-349 in D. E. Toweill, and J. W. Thomas,
editors. North American elk: ecology and management. Smithsonian Institution
Press, Washington, D. C., USA.
25
Cook, J. G., L. L. Irwin, L. D. Bryant, R. A. Riggs and J. W. Thomas. 1998. Relations of
forest cover and condition of elk: a test of the thermal cover hypothesis in summer
and winter. Wildlife Monographs 141:3-61.
Cooper, A. B., and J. J. Millspaugh. 1999. The application of discrete choice models to
wildlife resource selection studies. Ecology 80:566-575.
Croissant, Y. 2013. mlogit: multinomial logit model. R package version 0.2-4.
http://CRAN.R-project.org/package=mlogit. Accessed 5 Nov 2014.
Durner, G. M., D. C. Douglas, R. M. Nielson, S. C. Amstrup, T. L. McDonald, I. Stirling,
M. Mauritzen, E. W. Born, Ø. Wiig, E. DeWeaver, M. C. Serreze, S. E. Belikov,
M. M. Holland, J. Maslanik, J. Aars, D. A. Bailey, and A. E. Derocher. 2009.
Predicting 21st-century polar bear habitat distribution from global climate models.
Ecological Monographs 79:25-58.
Gelman, A., J. B. Carlin, H. S. Stern, D. B. Dunson, A. Vehtari, and D. B. Rubin. 2014.
Bayesian data analysis. Third edition. CRC Press, Boca Raton, Florida, USA.
Gilks, W. R., A. Thomas, and D. J. Spiegelhalter. 1994. A language and program for
complex Bayesian modelling. Journal of the Royal Statistical Society Series D
43:169-177.
Griffith, B., J. M. Scott, J. W. Carpenter, and C. Reed. 1989. Translocation as a species
conservation tool: status and strategy. Science 245:477-480.
Griffith, J. A., E. A. Martinko, and K. P. Price. 2000. Landscape structure analysis of
Kansas at three scales. Landscape and Urban Planning 52:45-61.
Hillard, E. M. 2013. Elk (Cervus elaphus L.) habitat selection in Great Smoky Mountains
National Park. Thesis, Western Carolina University, Cullowhee, USA.
26
Howlin, S., W. P. Erickson, and R. M. Nielson. 2004. A validation technique for
assessing predictive abilities of resource selection functions. Pages 40–51 in
Western Ecosystem Technologies, Inc., editor. Proceedings of the First
International Conference on Resource Selection Laramie, Wyoming, January 13–
15, 2003. Resource Selection Methods and Applications. Western, EcoSystems
Technology, Inc., Cheyenne, Wyoming, USA.
Irwin, L. L., and J. M. Peek. 1983. Elk habitat use relative to forest succession in Idaho.
Journal of Wildlife Management 47:664-672.
Kéry, M. 2010. Introduction to WinBUGS for ecologists. Elsevier, Burlington,
Massachusetts, USA.
Larkin, J. L., J. J. Cox, M. W. Wichrowski, M. R. Dzialak, and D. S. Maehr. 2004.
Influences on release-site fidelity of translocated elk. Restoration Ecology 12:97-
105.
Larkin, J. L., D. S. Maehr, J. J. Cox, D. C. Bolin, and M. W. Wichrowski. 2003.
Demographic characteristics of a reintroduced elk population in Kentucky.
Journal of Wildlife Management 67:467-476.
Lausch, A., and F. Herzog. 2002. Applicability of landscape metrics for the monitoring of
landscape change: issues of scale, resolution and interpretability. Ecological
Indicators 2:3-15.
Long, R. A., J. L. Rachlow, and J. G. Kie. 2009. Sex-specific responses of North
American elk to habitat manipulation. Journal of Mammalogy 90:423-432.
27
Lupardus, J. L., L. I. Muller, and J. L. Kindall. 2011. Seasonal forage availability and diet
for reintroduced elk in the Cumberland Mountains, Tennessee. Southeastern
Naturalist 10:53-74.
Lyon, L. J. 1979. Habitat effectiveness for elk as influenced by roads and cover. Journal
of Forestry 77:658-660.
Lyon, L. J., and A. G. Christensen. 2002. Elk and land management. Pages 556-581 in D.
E. Toweill, and J. W. Thomas, editors. North American elk: ecology and
management. Smithsonian Institution Press, Washington, D. C., USA.
Lyon, L. J., and C. E. Jensen. 1980. Management implications of elk and deer use of
clear-cuts in Montana. Journal of Wildlife Management 44:352-362.
Marcum, C. L. 1975. Summer-fall habitat selection and use by a western Montana elk
herd. Dissertation, University of Montana, Missoula, Montana, USA.
McClafferty, J. A. 2000. An assessment of the biological and socioeconomic feasibility
of elk restoration in Virginia. Thesis, Virginia Polytechnic Institute and State
University, Blacksburg, USA.
McGarigal, K., SA Cushman, and E Ene. 2012. FRAGSTATS v4: Spatial Pattern
Analysis Program for Categorical and Continuous Maps. Computer software
program produced by the authors at the University of Massachusetts, Amherst.
Available at the following web site:
http://www.umass.edu/landeco/research/fragstats/fragstats.html.
McCorquodale S. M., K. J. Raedeke, and R. D. Taber. 1986. Elk habitat use patterns in
the shrub-steppe of Washington. Journal of Wildlife Management 50:664-669.
28
Meinert, D., T. Nigh, and J. Kabrick. 1997. Landforms, geology, and soils of the MOFEP
study area. Pages 56-68 in S. R. Shifley and B. L. Brookshire, editors.
Proceedings of the Missouri Ozark Forest Ecosystem Project Symposium: An
Experimental Approach to Landscape Research. General Technical Report NC-
193. U. S. Department of Agriculture, Forest Service, North Central Research
Station, St. Paul, Minnesota, USA.
Merrill, E. H. 1991. Thermal constraints on use of cover types and activity time of elk.
Applied Animal Behaviour Science 29:251-267.
Millspaugh, J. J., K. J. Raedeke, G. C. Brundige, and C. C. Willmott. 1998. Summer bed
sites of elk (Cervus elaphus) in the Black Hills, South Dakota: considerations for
thermal cover management. American Midland Naturalist 139:133-140.
Missouri Department of Conservation [MDC]. 2010. Elk Restoration in Missouri.
Missouri Department of Conservation, Jefferson City, USA.
Missouri Department of Conservation [MDC]. 2013. Peck Ranch Conservation Area.
<http://mdc4.mdc.mo.gov/applications/moatlas/AreaSummaryPage.aspx?txtAreaI
D=5203>. Accessed 20 Mar 2013.
Murrow, J. L. 2007. An experimental release of elk into Great Smoky Mountains
National Park. Thesis, University of Tennessee, Knoxville, USA.
Peled, T. 2010. Reintroductions as an ecosystem restoration tool: a case study of
reintroduced ungulates as vectors for seed-dispersal. Thesis, Ben-Gurion
University of the Negev, Beer-Sheeva, Israel.
29
Popp, J. N., D. N. C. McGeachy, and J. Hamr. 2013. Elk (Cervus elaphus) seasonal
habitat selection in a heterogeneous forest structure. International Journal of
Forestry Research 2013:1-7.
Popp, J. N., T. Toman, F. F. Mallory, and J. Hamr. 2014. A century of elk restoration in
eastern North America. Restoration Ecology 22:723-730.
Ripple, W. J., and R. J. Beschta. 2004. Wolves and the ecology of fear: can predation risk
structure ecosystems. Bioscience 54:755-766.
Rocky Mountain Elk Foundation (RMEF) and Southern & Eastern Kentucky Tourism
Development Association. 2007. Study of the elk and wildlife viewing potential
for southern and eastern Kentucky. 163 pp.
Rota, C. T., M. A. Rumble, J. J. Millspaugh, C. P. Lehman, and D. C. Kesler. 2014.
Space-use and habitat associations of black-backed woodpeckers (Picoides
arcticus) occupying recently disturbed forests in the Black Hills, South Dakota.
Forest Ecology and Management 313:161-168.
Royall, R.M. 1997. Statistical evidence: a likelihood paradigm. Chapman and Hall, New
York, New York, USA.
Rumble, M. A., L. Benkobi, and R. S. Gamo. 2007. A different time and place test of
ArcHSI: a spatially explicit habitat model for elk in the Black Hills. Research
Paper RMRS-RP-64. U.S. Department of Agriculture, Forest Service, Rocky
Mountain Research Station, Fort Collins, Colorado, USA.
Schneider, J., D. S. Maehr, K. J. Alexy, J. J. Cox, J. L. Larkin, and B. C. Reeder. 2006.
Food habits of reintroduced elk in southeastern Kentucky. Southeastern Naturalist
5:535-546.
30
Shifley, S. R., C. D. Rittenhouse, and J. J. Millspaugh. 2009. Validation of landscape-
scale decision support models that predict vegetation and wildlife dynamics.
Pages 415-448 in J. J. Millspaugh, and Frank R. Thompson, editors. Models for
Planning Wildlife Conservation on Large Landscapes. Elsevier/Academic Press,
Burlington, Massachusetts, USA.
Skolvin, J. M., L. D. Bryant, and P. J. Edgerton. 1989. Timber harvest affects elk
distribution in the Blue Mountains of Oregon. USDA Forest Service Research
Paper PNW-RP-415, Portland, Oregon, USA.
Skolvin, J. M., P. Zager, and B. K. Johnson. 2002. Elk habitat selection and evaluation.
Pages 530-555 in D. E. Toweill, and J. W. Thomas, editors. North American elk:
ecology and management. Smithsonian Institution Press, Washington, D. C.,
USA.
Telesco, R. L., F. T. Van Manen, J. D. Clark, and M. E. Cartwright. 2007. Identifying
sites for elk restoration in Arkansas. Journal of Wildlife Management 71:1393-
1403.
Thomas, D. L., D. Johnson, and B. Griffith. 2006. A Bayesian random effects discrete-
choice model for resource selection: population-level selection inference. Journal
of Wildlife Management 70:404-412.
Tousignant, J. 2011. The Current River Hills of Southern Missouri: A CCPI EQIP
Proposal. Missouri Department of Conservation, Jefferson City, USA.
Unsworth, J. W., L. Kuck, E. O. Garton and B. R. Butterfield. 1998. Elk habitat selection
on Clearwater National Forest, Idaho. Journal of Wildlife Management 62:1255-
1263.
31
Van Deelen, T. R., L. B. McKinney, M. G. Joselyn and J. E. Buhnerkempe. 1997. Can
we restore elk to southern Illinois? The use of existing digital land-cover data to
evaluate potential habitat. Wildlife Society Bulletin 25:886-894.
Walter, W. D. 2006. Ecology of a colonizing population of Rocky Mountain elk (Cervus
elaphus). Dissertation, Oklahoma State University, Stillwater, USA.
Wichrowski, M. W., D. S. Maehr, J. L. Larkin, J. J. Cox and M. P. Olsson. 2005.
Activities and movements of reintroduced elk in southeastern Kentucky.
Southeastern Naturalist 4:365-374.
Wolf, C. M., B. Griffith, C. Reed, and S. A. Temple. 1996. Avian and mammalian
translocations: update and reanalysis of 1987 survey data. Conservation Biology
10:1142-1154.
Wolff, J. O., and T. Van Horn. 2003. Vigilance and foraging patterns of American elk
during the rut in habitats with and without predators. Canadian Journal of
Zoology 81:266–271.
Xu, M., S. C. Saunders, and J. Chen. 1997. Analysis of landscape structure in the
southeastern Missouri Ozarks. Pages 41-55 in S. R. Shifley and B. L. Brookshire,
editors. Proceedings of the Missouri Ozark Forest Ecosystem Project Symposium:
An Experimental Approach to Landscape Research. General Technical Report
NC-193. U. S. Department of Agriculture, Forest Service, North Central Research
Station, St. Paul, Minnesota, USA.
32
TABLES
Table 1. Covariates used to model broad-scale elk resource selection.
Covariate Variable type Definition/measurement
Land cover type Categorical
(8a)
9 vegetation classes (see table 2) with “other” being
the reference type
% Canopy cover Continuous 0-100% of 30m cell covered with tree canopy
Time since burn Continuous Number of years since area was last burned;
updated each spring; 0-70 years
Aspect Continuous from 0° (flat) to 360° (due north)
Distance to edge Continuous Edge-the distinct transition from forest or woodland
to open land (savanna, shrub-land, grassland, or
forage opening); distance in meters from point to
edge
Interspersion and
juxtaposition index
(IJI)
Continuous Measurement of the distribution of habitat patches;
from least (0) to most (100) evenly distributed
Slope Continuous From 0% (flat) to 100% (45°)
Distance to road Continuous (3) Distance in meters from point to nearest road by
road type
Road density Continuous Length of public roads in km within a 550m radius
a represents the number of variables (if >1) included in the model for that particular
covariate (total variables = 18).
33
Table 2. Definitions of the 6 land cover types used in broad-scale elk resource selection
analysis.
Land Cover Type Definition
Closed canopy forest Trees make up >80% of canopy cover
Woodlands Trees make up 50-80% of canopy cover
Savannas Sparse woodlands with 20-50% of canopy cover
Shrub-lands Majority of vegetation made up of deciduous
shrubs and young trees; mostly regenerating
clear-cuts
Glades Natural opening made up of mostly herbaceous
vegetation, some stunted trees and shrubs, and
exposed bedrock
Warm-season grassland Grassland made up of mostly native warm-
season grasses
Cool-season grassland mostly pasture managed for livestock and hay
production made up of mostly exotic grasses
Forage openings Open lands maintained by MDC and NPS and
planted with various grasses, grains, and
legumes for wildlife
Other Includes urban, barren land, crop land, and
water
34
FIGURES
Figure 1. Missouri elk restoration zone map depicting land ownership. The Missouri
Department of Conservation and Missouri Spatial Data Information Service (MSDIS)
provided these GIS layers.
35
Figure 2. Map of Peck Ranch Conservation Area. GIS layers were provided by the
Missouri Department of Conservation, Missouri Spatial Data Information Service
(MSDIS), and the U.S. Census Bureau.
36
Figure 3. Histogram demonstrating the distribution of GPS locations of elk by time of
day.
37
Figure 4. Radius of available habitat method. In this method, the length of the radius is
calculated according to the movement potential of an elk using the equation above, where
a is the monthly mean hourly movement rate, b is the standard error of the movement rate
and h is the number of hours between locations (Durner et al. 2009). The radius is then
used to create a circle with the previous use location as the center. Five points are then
randomly projected within the circle to represent the habitat available to the elk when it
selected the current use location.
Previous Use Location
Current Use Location
Available Locations (a + 2b)h
38
Figure 5. Estimated relative probability (± 95% credible intervals) of an elk using a sample unit based on land cover class. This
figure assumes a choice set of 9 alternatives, 1 for each land cover class including the reference class (“other”). W-S grass, C-S grass,
and F.O. represent warm-season grassland, cool-season grassland, and forage opening, respectively.
39
Figure 6. Estimated relative probability (± 95% credible intervals) of an elk using a
sample unit based on canopy cover, interspersion juxtaposition index, slope, and years
since prescribed fire. In these figures, elk are assumed to be presented with two choice
sets; one represented by the value of the solid line and the other represented by the values
of the x-axis.
40
Figure 7. Estimated relative probability (± 95% credible intervals) of an elk using a
sample unit based on distance to nearest two-track road and distance to nearest paved
road. In these figures, elk are assumed to be presented with two choice sets; one
represented by the value of the solid line and the other represented by the values of the x-
axis.
41
Figure 8. Estimated relative probability (± 95% credible intervals) of an elk using a
sample unit based on distance to nearest public gravel road, road density, aspect, and
distance to edge. In these figures, elk are assumed to be presented with two choice sets;
one represented by the value of the solid line and the other represented by the values of
the x-axis.
42
Figure 9. Estimated relative probability (± 95% credible intervals) of an elk using a sample unit based on land cover class during
summer, fall, winter, and spring, respectively. These figures assume a choice set of 9 alternatives, 1 for each land cover class
including the reference class (“other”). Wood, Sav, Shrub, WSG, CSG, and F.O. represent woodland, savanna, shrub-land, warm-
season grassland, cool-season grassland, and forage opening, respectively.
43
Figure 10. Estimated relative probability (± 95% credible intervals) of an elk using a
sample unit during each season as a function of distance to the nearest wooded edge. In
these figures, elk are assumed to be presented with two choice sets; one represented by
the value of the solid line and the other represented by the values of the x-axis.
44
Figure 11. Estimated relative probability (± 95% credible intervals) of an elk using a
sample unit during each season as a function of road density. In these figures, elk are
assumed to be presented with two choice sets; one represented by the value of the solid
line and the other represented by the values of the x-axis.
45
Figure 12. Estimated relative probability (± 95% credible intervals) of an elk using a
sample unit during each season as a function of the interspersion juxtaposition index (IJI).
In these figures, elk are assumed to be presented with two choice sets; one represented by
the value of the solid line and the other represented by the values of the x-axis. IJI
increases as the interspersion of different land cover types increases.
46
Figure 13. Estimated relative probability (± 95% credible intervals) of a female or male
elk using a sample unit as a function of distance to two-track road and road density. In
these figures, elk are assumed to be presented with two choice sets; one represented by
the value of the solid line and the other represented by the values of the x-axis.
47
Chapter II: Food Habits and Diet Selection of a Recently Reintroduced Elk
Population in Missouri
ABSTRACT
Since elk (Cervus elaphus) were extirpated from eastern North America, they
have been successfully reintroduced in 8 eastern states and 1 Canadian province and have
recently been reintroduced in Missouri and Virginia. Although some aspects of the
ecology of reintroduced elk populations have been studied, resource needs are not well
understood. The purpose of our study was to determine the seasonal diet selection of
recently reintroduced elk in Missouri by comparing use (diet composition) ranks with
forage availability ranks of 12 forage classes. We measured diet composition through the
microhistological analysis of feces. We collected feces from June 2011 through February
2013 by randomly selecting an individual elk 3 times per week and searching for fresh
fecal samples near the most recent location from its GPS collar. We determined forage
availability through vegetation sampling at 201 random points stratified by vegetation
type. Elk selected grains and cool-season grasses over all other forage classes across all
seasons and years except during summer 2012 and fall 2012. Legumes were the most
highly consumed forage class for all seasons and years. Native forbs were ranked as
highly selected during the 2012 drought (summer and fall), but were not selected by elk
during 2011. Approximately half (44.6%) of the annual elk diet was composed of plant
species cultivated in wildlife forage openings (fields cultivated to provide forage for
wildlife). Clover (Trifolium spp.), alfalfa (Medicago sativa), and common lespedeza
(Kummerowia spp.) were the most highly consumed species by elk. In forested
48
dominated landscapes, we recommend the creation of forage openings composed of cool-
season grasses and legumes to increase forage availability for elk. Management for
natural open lands (e.g. glades) composed of native forbs and grasses may provide an
alternative food source when food is scarce.
INTRODUCTION
Nutrition is a critical component of wildlife ecology, because it determines the
resources available for growth, reproduction, and survival (Cook 2002). Understanding
the food habits of large herbivores is particularly important both to ensure they have
sufficient forage on the landscape and to determine their potential impacts on vegetation
structure and composition. Armed with this information, managers can manipulate the
landscape to increase forage availability or implement population management to
decrease the effects of herbivory. Furthermore, understanding food habits is crucial for
declining or establishing populations of herbivores because forage availability can limit
population growth.
Elk (Cervus elaphus) were extirpated from eastern North America by the end of
the 19th century. Since then, 8 states and 1 Canadian province have successfully restored
elk (Arkansas, Kentucky, Michigan, Minnesota, North Carolina, Pennsylvania,
Tennessee, Wisconsin, Ontario; Popp et al. 2014), and 2 states have recently reintroduced
elk (Missouri in 2011 and Virginia in 2012). Missouri made their final release of elk on
June 7, 2013. Reintroducing elk has been popular in many states, because it has the
potential to provide ecological and economic benefits (Ripple and Beschta 2004, RMEF
and Southern & Eastern Kentucky Tourism Development Association 2007, Peled 2010).
49
Habitat management is crucial for these recently reintroduced populations both to
provide the appropriate resources to facilitate establishment and to minimize human
conflicts. One element necessary for sound habitat management is an understanding of
food habits. The food habits of western elk populations have been thoroughly described
with compilations of studies for Rocky Mountain elk (Cervus elaphus nelsoni) (Kufeld
1973), Roosevelt elk (C. elaphus roosevelti) (Jenkins and Starkey 1991), and North
American elk in general (C. elaphus) (Cook 2002). However, food habits are not well
understood in eastern populations of elk. Murphy (1963) surveyed the food habits of a
captive elk population in Missouri, but this information was collected during the fall of a
single year as the population was being removed. More recently, food habit studies of
eastern populations were conducted in Kentucky (Schneider et al. 2006), North Carolina
(Murrow 2007), and Tennessee (Lupardus et al. 2011). The studies in Tennessee and
Kentucky occurred in deciduous forests similar to that of the elk restoration zone in
Missouri, except the landscapes had been largely altered by coal mining (Schneider et al.
2006, Lupardus et al. 2011). In those study areas, many relatively large open lands
(>100ha) were created during mine reclamation (Larkin et al. 2004), which do not exist in
Peck Ranch Conservation Area where the elk in Missouri currently reside. Furthermore,
with the Missouri Department of Conservation’s (MDC) investment in managing forage
openings (fields cultivated to provide forage for wildlife) and woodlands, particularly on
Peck Ranch, we need to understand whether these vegetation types provide important
food resources to elk. Open lands (areas with essentially 0% canopy cover including
glades, pastures, and forage openings) are relatively small (average =1 ha) in this area
with the majority of the Ozark landscape being forested, so elk may depend on other,
50
more abundant forages such as woody browse and hard mast (Cook 2002, Lupardus et al.
2011). Also, determining which forages elk select, rather than diet composition alone,
will provide a clearer picture of the importance of certain foods. Thus, we must
determine what plants elk select in Missouri to manage for the production of these foods.
Elk are considered an intermediate feeder meaning their diet usually consists of a
mixture of grasses and woody browse with neither being dominant (Cook 2002, Keller
2011). However, because elk are found in such a wide variety of habitats, including
grasslands, shrub-lands, and coniferous and deciduous forests, their diet is extremely
variable (Cook 2002). The food habits studies from Tennessee (Lupardus et al. 2011)
and Kentucky (Schneider et al. 2006) offer some insight into what foods might be
important to eastern populations of elk. Both studies agree the most dominant forage
classes in elk diets are forbs in summer, woody browse (including hard mast) during fall,
and grasses during winter. During spring, elk in Tennessee consumed mostly grasses
while elk in Kentucky consumed mostly woody browse. These results indicate elk food
habits are predominantly determined by what is present and of highest quality on the
landscape each season (i.e., availability and phenology; Jenkins and Starkey 1991, Cook
2002).
Our objective was to determine food habits of the reintroduced elk population in
Missouri. We determined diet composition and diet selection of elk seasonally for 12
forage classes. According to the availability and phenology of plants in Missouri, we
predicted elk would select forbs during summer, woody browse and acorns during fall,
and grasses during winter and spring.
51
STUDY AREA
In 2010, MDC designated an 896 square-kilometer elk restoration zone in
Shannon, Reynolds, and Carter counties in southeast Missouri (Figure 1) (MDC 2010).
This area was selected based on its potential for elk habitat, a high percentage of public
land ownership, few row crops, and limited public roads (MDC 2010). The elk
restoration zone is located in the Missouri Ozarks and is composed mostly of oak-hickory
and oak-pine forests and woodlands with the landscape being 93% forested, 5% open,
and 0.1% cropland (Xu et al. 1997, MDC 2010). The majority of the land (79%) is open
to the public with 49% in public ownership and managed by MDC, the United States
Forest Service, and the National Park Service with the remaining 30% owned and
managed by The Nature Conservancy and the L-A-D Foundation (a local private
organization; MDC 2010).
Peck Ranch Conservation Area near Van Buren, Missouri was the center for the
elk reintroductions, and was where most of the elk in Missouri were located during this
study (Figure 2). This is a 9,327 hectare parcel managed by MDC (MDC 2013). A large
portion of this conservation area (4,283 hectares) is maintained as a wildlife refuge where
hunting is only allowed during three managed deer (Odocoileus virginianus) hunts
annually. The landscape of Peck Ranch is made up of a mosaic of ridges and valleys.
Elevation on most ridges in Peck Ranch ranges from 275 to 300 meters with a maximum
elevation of 411 meters on Stegall Mountain (MDC 2013). This area contains soils that
range from deep and loamy to shallow and rocky with dolomite, sandstone, or igneous
rhyolite substrates (Meinert et al. 1997).
52
Like the rest of the restoration zone, Peck Ranch is composed primarily of oak-
dominated forest and woodland (91%) with 9% open lands in glades and forage openings
(MDC 2013). The most common overstory tree species in this area are white oak
(Quercus alba), black oak (Q. velutina), scarlet oak (Q. coccinnea), post oak (Quercus
stellata), and shortleaf pine (Pinus echinata) (Kabrick et al. 2002). Common understory
species include Nuttall’s trefoil (Desmodium nuttalli), Virginia creeper (Parthenocissus
quinquefolia), hog peanut (Amphicarpa bracteata), summer grape (Vitis aestivalis), low-
bush blueberry (Vaccinium pallidum), and black-edged sedge (Carex nigromarginata)
(Chen et al. 2002). Even and uneven aged forest management are utilized in Peck Ranch
including management activities such as small clearcuts (<30 acres), single-tree selection
and group-selection cuts, and thinning. Prescribed fire is also an important component of
forest management in the Missouri Ozarks occurring on a 3 year rotation on relatively
large burn units (average 300 ha). On average, 15% of Peck Ranch is burned each year.
MDC has a management goal to increase, restore, and maintain woodlands,
glades, and forage openings on the landscape for wildlife including elk (MDC 2010).
Woodlands are believed to provide more forage for elk than closed canopy forests,
because their relatively open canopies allow for the growth of more understory
vegetation. Glades are open lands with shallow rocky soils and sparse vegetation. These
open areas dot the landscape and have a diversity of herbaceous plant species with some
small shrubs and stunted trees. Peck Ranch has over 100 forage openings ranging in size
from 0.05ha to 8ha. Most forage openings are planted with one of three grains, wheat
(Triticum aestivum), barley (Hordeum vulgare), or rye (Secale cereal), each fall. The
predominant cultivated cool-season grass in forage openings is orchardgrass (Dactylis
53
glomerata) and the established legumes are clover (Trifolium spp.) and alfalfa (Medicago
sativa). Some forage openings also include warm-season grasses such as big bluestem
(Andropogon gerardii), indiangrass (Sorghastrum nutans), and eastern gammagrass
(Tripsacum dactyloides). MDC invested greatly to develop and maintain these forage
openings to provide additional forage for the elk.
The mean annual temperatures for our study area for 2011, 2012, and 2013 were
12.3, 13.7, and 14.9 ̊ C, respectively, with maximum temperatures occurring in July (30,
35, and 37 ̊C) and minimum temperatures occurring in January (-4, -5, and -7 ̊ C) (Prism
Climate Group, www.prism.oregonstate.edu, accessed 23 Sept. 2014). Most
precipitation in this area is in the form of rainfall. A severe drought occurred throughout
the Midwestern United States including our study area during the summer of 2012.
Limited rainfall was most evident during June. June precipitation in our study area was
slightly less than the 30-year (1981-2010) average (97mm) during 2011 (71mm) and
slightly greater than average in 2013 (121mm; Prism Climate Group 2014). June
precipitation in 2012 was less than half of the 30-year average (47mm; 48% of average
precipitation; Prism Climate Group 2014 ).
METHODS
Reintroduction
During the summers of 2011, 2012, and 2013, a total of 106 adult elk were
released in the Missouri elk restoration zone. These animals were captured in eastern
Kentucky, translocated to Missouri, and held in holding pens for up to 4 weeks to
54
facilitate a soft-release. Each animal > 1 year of age was fitted with a GPS collar
(RASSL custom 3D cell collar, North Star Science and Technology, LLC, King George,
VA, or G2110E Iridium/GPS series model, Advanced Telemetry Systems, Insanti, MN)
prior to being released. These collars recorded on-board locations every 2-2.5 hours and
uploaded a location every 4-5 hours that was accessible via the internet. All animal
activities were approved by the University of Missouri Animal Care and Use Committee
(Protocol 6909).
Fecal Sample Collection and Analysis
We used microhistological analysis of feces to determine the percent diet
composition for each season staring in July 2011. We began collected fecal samples 2
ways. Three times per week, we randomly selected an individual elk and determined its
location according to recently (within the past 12 hours) uploaded GPS coordinates. We
uploaded the coordinates to a handheld GPS unit, navigated to the location, searched the
general area, and collected 1-2 fresh (wet and dark) samples. We assumed that most
animals were with a group and we avoided collecting 2 samples within 50m of one
another. For every sample collected, we recorded the date and location (latitude and
longitude). We also collected fecal samples opportunistically while conducting other
research activities throughout the area (vegetation surveys, transects for deer fecal
samples, snail collection, etc.). We placed all samples in a freezer as soon as possible to
preserve freshness. We set a goal to collect at least 5 total fecal samples per week. We
then combined samples into two-week composite samples with 2 fecal pellets from each
55
individual sample, which is common in diet studies (Jenks et al. 1989, Alipayo et al.
1992, Schneider et al. 2006, Keller 2011).
We submitted all composite samples to the Washington State University Wildlife
Habitat and Nutrition Lab (WSU Habitat and Nutrition Lab; Pullman, Washington) for
microhistological analysis. There, lab technicians prepared 6 slides for each composite
fecal sample and used microhistological analysis to determine percent diet composition to
genus or species level of identification. Technicians randomly selected 25 views per
microscope slide with 6 slides per sample for a total of 150 views per two-week
composite sample. For each view, they used a microscope at 100X magnification with a
10x10 square grid in the eyepiece to determine the relative cover of each identifiable
plant fragment. The technicians then recorded the total area covered by each genus or
species, divided it by the total area covered by all species in the sample, and multiplied
by 100 to determine the percent diet composition for each species in the sample.
Correction Factors
Differential digestibility of plant species can be a source of bias in
microhistological analysis of feces (Pulliam and Nelson 1979, Vavra and Holechek
1980). To adjust for this bias, we determined correction factors for each season based on
a technique developed by the WSU Habitat and Nutrition Lab (B. B. Davitt, Washington
State University, unpublished data) which was adapted from Pulliam and Nelson (1979).
First, we determined the plant species that were dominant (≥ 5% diet composition) in the
seasonal elk diets that we observed from summer 2011-spring 2012. We collected
samples of each of these species during the appropriate season from winter 2013-fall
56
2014 to account for seasonal differences in digestibility (Pulliam and Nelson 1979). We
dried these samples in a drying oven at 50̊ C for 48 hours, then ground them into 2mm
particles using a Wiley mill (Swedesboro, NJ), and then into 1mm particles using a Foss
sample mill (Tecator Cyclotech 1093 sample mill, FOSS, Eden Prairie, MN).
We used the plant samples to reconstruct the observed elk diets for each season by
combining the appropriate proportions of 3 broad forage classes (forbs, grasses/sedges,
and woody browse; B. B. Davitt, unpublished data). Within each forage class, we only
included plant material from the dominant species (≥ 5% of diet) for that season. We
also included a standard plant (pine needles) in all of the reconstructed diets at a rate of
5% making the total reconstructed diet percentage equal to 105% (Pulliam and Nelson
1979). We then performed in vitro digestion on each reconstructed diet using a Daisy
incubator (120v Daisy II Incubator, Ankom Technology, Macedon, NY) and fresh cattle
(Bos taurus) rumen fluid. We followed the standard protocol for this incubator, which
includes preparation of plant samples and 2 buffer solutions, procedure for handling fresh
rumen fluid, and combining these components for in vitro digestion. We first took 2,
0.5g samples of each reconstructed diet and placed each sample in a filter bag. We then
placed the filter bags in a glass jar containing rumen fluid and the buffer solutions. We
placed the jar in the incubator and performed in vitro digestion for 48 hours.
After the digestion, we bleached, dyed, and mounted the digested material from
each diet on 6 microscope slides following the protocol used by the WSU Habitat and
Nutrition Lab. We determined the frequency of occurrence of plant fragments within
each forage class as a measure of diet composition (Keller 2011). To accomplish this, we
randomly selected 25 views on each microscope slide using a random number generator.
57
At each random view, we recorded the forage class of the nearest identifiable plant
fragment to the center of the view. We identified plant fragments as forbs,
grasses/sedges, woody browse, or standard plant based upon shape and size of cell walls,
presence of trichomes, and overall arrangement of cells. We made reference slides of
plant species to assist in identification. We calculated the frequency of each forage class
in the digested diet as the number of views containing that forage class divided by the
total number of views (150). We calculated correction factors (CF) for each forage class
during each season as:
𝐶𝐹 = (𝑓𝑠 × 𝑑𝑓) (𝑓𝑓 × 𝑑𝑠)⁄
where fs is the frequency of the standard plant after digestion, df is the known % diet
composition of the forage class, ff is frequency of the forage class after digestion, and ds
is known % diet composition of the standard plant (5%; Pulliam and Nelson 1979). We
then multiplied the % diet composition of each plant species in each two-week sample by
the appropriate correction factor and recalculated diet composition. The correction
factors are all relative to the standard plant (Pulliam and Nelson 1979). Therefore, they
should only be interpreted after being applied to the diet composition. Forage classes that
were overestimated in the original diet show a decrease in diet composition while classes
that were underestimated show an increase in diet composition after the application of
correction factors. We were unable to develop a correction factor for the “other” forage
class which contained ferns, lichens, and moss, because these forages are quite different
from one another. We instead calculated a correction factor for this forage class using the
equation above assuming no change after digestion. We also only calculated a correction
factor for hard mast during fall, because that is when hard mast is most abundant.
58
Forage Availability
Diet composition reveals what elk are consuming, but to determine what plants
elk select we measured availability of forage plants. To determine forage availability, we
sampled herbaceous and low-growing woody vegetation, and estimated hard mast
availability. Because the majority of fecal samples are collected within Peck Ranch
Conservation Area, we considered it the study area for sampling purposes. We used a
stratified random sampling approach to determine the placement of our vegetation points
similar to Lupardus et al. (2011). First, we determined the percentage of area covered by
5 main land cover types (closed canopy forest, dense woodland, sparse
woodland/savanna, shrub-lands, and open lands) within Peck Ranch using GIS layers
developed by the Missouri Resource Assessment Partnership and MDC. We then
randomly selected a number of points in each land cover type according to their percent
cover within the study area using ArcInfo 10 (Environment Systems Research Institute,
Inc., Redlands, CA).
To determine the total number of points needed to adequately sample the study
area, we used the standard formula for sample size calculation for proportions of an
infinite population:
𝑛 =𝑍2(𝑝)(1 − 𝑝)
𝐸2.
where Z is the standardized Z-value for the selected confidence level (1.96=95%
confidence), p is an estimate of the proportion of a certain trait in the population (percent
59
cover), and E is the desired level of precision or sampling error (10%). For the
proportion, we used the percent cover of the most abundant plant species in the Missouri
Ozarks plus one standard deviation from Grabner (2000). We determined that
approximately 82 sampling points were adequate to obtain a representative sample with
10% sampling error.
At each sampling point, we placed a 30m transect in a randomly selected cardinal
direction (north, south, east, or west) starting at the point using a meter tape. We then
placed 0.1m2 (20x50cm) quadrats perpendicular to the transect to sample understory
vegetation (≤1.5m in height) on alternating sides every meter (1-30) for a total of 30
quadrats per transect (Daubenmire 1959, Higgins et al. 2012) (Figure 3). We visually
estimated percent cover of each herbaceous (non-woody) plant species within each
quadrat, including plant parts overlapping from the outside, using the 6 cover classes
described by Daubenmire (1959): class 1: >0-5%, class 2: 6-25%, class 3: 26-50%, class
4: 51-75%, class 5:76-95%, class 6:96-100%. For these cover classes, we used the mid-
point of the class as the estimated value in our calculations (e.g. class 1=2.5%).
Because a larger area is usually required to adequately sample woody browse
(Higgins et al. 2012), we visually estimated the cover of each woody plant species with
vegetation <1.8m in height within a 25m2 quadrat (Keller 2011) at each sampling point.
We estimated cover of each species in woody browse plots to the nearest 1%, except in
plots with very dense vegetation where we estimated cover to the nearest 5%. We laid
out this quadrat in the compass direction opposite the transect described above.
According to our preliminary observations, elk used forage openings heavily.
Therefore, in addition to the points chosen above, we randomly selected 6 forage
60
openings within Peck Ranch, 3 cool-season and 3 mixed-season (contain cool and warm-
season grasses) to sample. We chose 6 forage openings because it was the number of
points that was proportional to the area covered by forage openings in the study area, thus
maintaining our stratified random sampling strategy. Vegetation sampling of forage
openings was performed separate from this food habits study with a slightly different
protocol than the rest of the sampling points (B.J. Keller, University of Missouri,
unpublished data). This separate study sampled 74 forage openings within Peck Ranch
Conservation Area. Within each forage opening, we placed the sampling point in
approximately the center. We then ran 4 transects, one in each cardinal direction (north,
south, east, and west), from the central point to the edge of the forage opening using a
meter tape. We placed a 0.1m2 (20x50cm) quadrat perpendicular to the tape every 10m
on alternating sides along each transect until we reached the edge of the forage opening.
We also sampled one quadrat at the central point. The number of quadrats sampled for
each forage opening varied depending on the size of the forage opening. We visually
estimated percent cover of each herbaceous plant species within each quadrat as
described above. We did not sample woody vegetation for the forage opening sampling
points. We sampled each forage opening during each season from fall 2012 through
summer 2014. We randomly selected 6 forage openings (3 cool-season, 3 mixed-season)
for each season/year of diet data (eg. summer 2011, summer 2012, fall 2011, fall 2012,
etc.; see season dates below) to capture as much of the variability among forage openings
as possible in our sample.
In oak-dominated forests, acorns are known to be an important food source in fall
and winter for many wildlife species, especially white-tailed deer (McShea and Schwede
61
1993). Acorn production in Peck Ranch had been estimated since 1993 as a part of the
Missouri Ozark Forest Ecosystem Project (MOFEP; Vangilder 1997). This study
estimated the mean number of sound mature acorns produced using acorn traps on 130
plots throughout the Missouri Ozark region. We used these data from MOFEP to
estimate acorn availability. Because acorn production varies greatly, we used only the
annual estimates for 2011 and 2012 from 44 plots on Peck Ranch. Because of variation
in acorn production between and even within forest stands, variation in acorn size, and
other confounding factors, we decided that it was not reasonable to estimate hard mast
availability in each 0.1m2 quadrat and therefore excluded hard mast from our diet
selection analyses. Instead, we compared acorn use with overall acorn availability each
year to determine if elk use changed with availability. To accomplish this, we calculated
the long-term mean number of acorns per hectare from 1993-2012 for Peck Ranch from
MOFEP. We then calculated the annual mean number of acorns per hectare for 2011 and
2012. We determined whether acorn production for each year was approximately below
average, average, or above average by subtracting the annual mean from the overall
mean. Finally, we compared the proportion of acorns in each fall and winter elk diet with
the appropriate annual acorn production estimate to determine if they vary together.
We collected forage availability data along 201 transects and within 201 woody
browse quadrats, not including forage openings or acorn plots. We determined the
overall mean percent cover for all species observed for each season (see season dates
below) including vegetation in forage openings. Because of logistical constraints, we
only sampled vegetation during summer 2013 (May 22-Septemeber 26) and 2014 (May
19-August 22) and sampled each point only once. However, this approach accurately
62
represented of grass, forb, and browse availability because this time period included part
or all of the growing season for these plant types. We assumed that forage availability
did not change between years. Because some plant species are absent or not palatable
during fall, winter, and spring, we excluded them from forage availability estimates for
these seasons. We decided which species to exclude from each season by consulting two
botanists with expertise in plants of the Missouri Ozarks (Susan Farrington and Elizabeth
Olson, Missouri Department of Conservation, personal communication).
Diet Selection Analysis
We compared diet composition (use) and forage availability to determine which
forages elk selected seasonally. We divided the seasons according to the following dates:
spring (March 1- May 15), summer (May 16- September 15), fall (September 16-
November 30), and winter (December 1-February 29) and analyzed each season of each
year separately because we anticipated potential differences in selection between years
due to the 2012 drought. We used the method developed by Johnson (1980) to determine
diet selection, because it not only determines which forages are selected, but also
provides a system to rank forage items based on selection (Norbury and Sanson 1992,
Tanentzap et al. 2009). Because it would not be practical to rank over 200 species of
plants, we grouped plants into 12 categories (cool-season grasses, warm-season grasses,
grains, sedges, weedy forbs, woodland forbs, legumes, other forbs, deciduous trees,
evergreen trees, shrubs and vines, other forage; see Appendix I for list of species). We
excluded unidentified grasses, forbs, and woody browse from the food habits analyses,
because we could not determine into which forage class they fit. Using Johnson’s (1980)
63
technique, for each two-week diet sample, we ordered the forage categories based on
mean percent cover (availability) in the study area and based on percent of the diet (use),
and then assigned each forage class 2 ranks (from greatest to least), one for availability
and one for use. We calculated the difference between their use and availability ranks
(use minus availability) and averaged the differences for each forage class across the
samples for that season. We ranked the forage classes according to these average
differences which created a selection ranking from least selected (1) to most selected
(12). Following Johnson’s (1980) protocol, we performed a standard Analysis of
Variance on the differences in use and availability ranks to determine if there were any
significant differences among the forage classes. When there were differences, we
performed a Waller and Duncan (1969) test to determine where the differences occurred.
We performed these calculations in R (R Version 3.0.1, www.R-project.org, accessed 17
May 2013) using the “agricolae” package (Mendiburu 2014).
We also calculated the selection ratio (Manly et al. 2002) for each forage class
during each season as:
𝑅 = 𝑢/𝑎
where u is the mean of the percent diet composition of the forage class for that season
and a is the mean of percent cover (availability) for that season. Selection ratios greater
than one indicate that % use is greater than % availability. These values are also directly
proportional to this relationship (e.g. R=2.5 means that use is 2.5 times greater than
availability). We used the selection ratios to further aid in interpretation.
64
RESULTS
Digestibility varied both among forage classes and seasons (Table 1). Woody
browse in general had low digestibility and was overestimated in the diet for all seasons
except during spring (Table 1). During summer, browse was overestimated in the diet by
2 times. Forbs were more digestible, resulting in an underestimation of their diet
composition across all seasons except fall (Table 1). Forbs were especially digestible
during spring and summer when they were underestimated in the diet by half. The
digestibility and correction factors of grasses varied across all seasons (Table 1). Grasses
was overestimated by almost 2 times in the diet during spring, but grasses was the most
digestible forage class during fall. As expected, acorns had relatively high digestibility,
resulting in their underestimation in the diet in fall (Table 1).
We collected 877 individual fecal samples (749 random, 128 opportunistic) from
July 10, 2011 to February 16, 2013, resulting in 42 composite samples. We observed 75
plant species in the elk diets and 207 species through forage availability sampling.
During summer, elk selected grains above all other forage classes in both 2011
and 2012 (Table 2). Cool-season grasses were also ranked as highly selected during
summer for both years (Table 2), but in 2012 their use was approximately equal to their
availability (selection ratio = 0.9). Similarly, weedy forbs were ranked as highly selected
for summer 2011 and 2012 (Table 2), but had low use relative to their availability in
2011. Elk selected the “other forbs” class in 2012, but not in 2011 (Table 2). Legumes
were used at a rate over 3 times their availability and made up over 50% of the summer
diet of elk during both 2011 and 2012 (Table 3). Common lespedeza (Kummerowia
spp.), clover (Trifolium spp.), and alfalfa (Medicago sativa) each made up over 10% of
65
the summer diet of elk for both 2011 and 2012. Other important forage plants (% diet
composition ≥ 5%) during this season included wild rye (Elymus spp.), rye (Hordeum
vulgare), wheat (Triticum aestivum), tick trefoil (Desmodium spp.) and native Lespedeza
spp.
In fall, grains were the most highly selected forage class in 2011 and 2012 (Table
2) and were consumed at a rate over 10 times their availability for both years. Cool-
season grasses were highly selected during fall 2011 (Table 2) with use 8.8 times their
availability, but were used at a rate less than their availability in 2012 (selection ratio =
0.8). Elk selected warm-season grasses in fall 2011, but not in 2012 (Table 2). Weedy
forbs and woodland forbs were used at rates 2.5 and 2.4 times their availability,
respectively in 2012, but were not highly selected by elk in fall 2011 (Table 2). Legumes
were not ranked highly for fall 2011 or 2012 (Table 2), but they made up ≥ 20% of the
fall elk diet during both years (Table 3). Hard mast, mostly oak (Quercus spp.) acorns,
made up 21.2% of the fall 2011 elk diet, but only made up 3.7% of the fall 2012 diet.
Acorn availability in 2011 was 2 times the mean across years for the study area while
acorn availability in 2012 was approximately half (0.6) of the mean across years,
therefore elk’s consumption of acorns was related to their availability. Clover made up
20.4% of the fall 2012 elk diet. Other important forage plants during fall included
orchard grass (Dactylis glomerata), dogwood (Cornus spp.), common lespedeza, alfalfa,
and barley (Hordeum spp.).
Grains and cool-season grasses were the most highly selected forage classes by
elk during winter 2011 and 2012 (Table 2). Elk consumed grains at a rate over 20 times
greater than their availability in winter 2011 (29.4) and 2012 (25.5). Legumes made up
66
over one quarter of the winter diet each year (Table 3) and were used at a rate over 6
times their availability for both years. Clover was the most commonly consumed forage
item during both winters with diet compositions of 17.7% and 16.6% for 2011 and 2012,
respectively. Common lespedeza, oak browse, ferns, wheat, orchard grass, alfalfa, sedges
(Carex spp.), and hard mast were also important forage plants during winter. We were
not able to estimate hard mast availability specifically for winter, but elk use of hard mast
during the winter of 2011 (7.5%) was much higher than 2012 (0.7%) which follows the
pattern of annual hard mast availability for these two years.
Once again, grains and cool-season grasses were ranked above all other forage
classes during spring even though they each made up less than 5% of the elk diet (Table
2). Legumes were also highly selected by elk and made up three-quarters (Table 3) of the
elk diet in spring 2012. Evergreen trees, sedges, and weedy forbs were not significantly
different from legumes, but they all had much lower selection ratios. Clover made up
almost half (45.7%) of the spring 2012 elk diet. Elk also consumed alfalfa at a relatively
high rate (19.0%) during spring. Other important forage plants during this season
included a native legume, purple prairie clover (Dalea purpurea), and oak browse.
Plant species cultivated by MDC and the National Park Service in forage
openings made up over one-quarter of the elk diet across all seasons and approximately
half of the annual diet (Table 4). During summer 2012 and winter 2011, cultivated
species made up approximately half of the elk diet and almost three-quarters of the elk
diet during spring 2012 (Table 4). Clover was consumed at the greatest rate out of all
cultivated plants and made up over 10% of the elk diet during every season except fall
2011 (Table 4).
67
DISCUSSION
A substantial portion of the Missouri elk diet during every season consisted of
species cultivated in forage openings. Grains and cool-season grasses were ranked as the
top 2 most selected forage classes across all seasons and years except fall 2012. Grains
are found exclusively in forage openings and cool-season grasses are found
predominantly in forage openings. Similarly, the most highly consumed forage class by
elk across all seasons was legumes. Legumes in the diet consisted mostly of clover,
alfalfa, and common lespedeza which were also found almost exclusively in forage
openings. Therefore, it is clear that elk in Missouri strongly selected forage cultivated in
forage openings. The importance of open lands which provide forage for elk has been
observed in both eastern and western North America (Cook 2002, Skolvin 2002,
Schneider et al. 2006, Lupardus et al. 2011). Some eastern populations of elk (Kentucky,
Tennessee, and Virginia) have access to large (>20ha) grassland areas created by coal
mine reclamation (Schneider et al. 2006, Lupardus et al. 2011). Maehr et al. (1999)
believed these open lands contributed to the successful establishment of the population in
Kentucky. Most other areas where elk have been reintroduced are dominated by forest
with few or no large open lands (e.g. North Carolina, Missouri, Wisconsin). The
landscape of the Missouri Ozarks is dominated by oak-hickory forests. MDC and the
National Park Service maintain over 200 small (average ≈ 2 ha) forage openings in the
elk restoration zone. Although these forage openings represent a small percentage of the
landscape overall, they appear to provide substantial forage for elk.
68
Elk use of legumes in Missouri throughout all seasons was much higher than
expected. Lupardus et al. (2011) observed that legumes made up approximately 20% of
elk diets during summer and fall. Similarly, legumes made up approximately 33% of the
annual elk diet in Kentucky (Schneider et al. 2006). In western populations, forbs only
made up substantial portions of the elk diet during summer (Kufeld 1973, Cook 2002).
In Missouri, legumes made up approximately half of the annual elk diet and over three
quarters of the spring diet. Legumes were also the most available forage class after
deciduous trees and shrubs and vines, which resulted in this forage class not being ranked
as highly selected during some seasons. However, the majority of the cover for the
legumes forage class was made up of native species such as tick trefoil (Desmodium
spp.), hog peanut (Amphicarpa bracteata), and native Lespedeza spp. The primary
legumes in the elk diet were clover, alfalfa, and common lespedeza which had relatively
low availability. Forbs also were the most digestible forage class for every season except
fall. The high palatability and protein content of these legumes makes them a valuable
resource for elk.
At the onset of this study, we predicted that elk would select forbs during
summer, browse during fall, and grasses during winter and spring based on the available
literature for eastern populations (Schneider et al. 2006, Lupardus et al. 2011). However,
grasses were highly selected year round, forbs (i.e. legumes) comprised the majority of
the diet composition across all seasons, and browse was never ranked as highly selected.
This resembles the food habits of many western populations where grasses and forbs are
more widely available (Kufeld 1973, Cook 2002). Elk diets in Tennessee and Kentucky
were more balanced with relatively equal proportions of grasses, forbs, and woody
69
browse (Schneider et al. 2006, Lupardus et al. 2011). As stated earlier the open lands in
Tennessee and Kentucky were predominantly reclaimed coal mines. While these open
lands are expansive, they are not intensively managed for forage production (i.e.
fertilized, limed, replanted regularly) like the forage openings in Missouri’s restoration
zone. Therefore, they may not provide enough palatable forage year round resulting in
the elk resorting to other food sources such as browse during fall and winter. Also,
Schneider et al. (2006) did not determine diet selection so they were not able to account
for the potentially high availability of autumn olive (Elaeagnus umbellata) and other
browse species in their study area. Deciduous trees in Missouri made up a substantial
portion of the fall and winter diets, but they also were the most available forage class
across all seasons. Therefore, deciduous trees were not highly selected by elk. Elk in
Missouri were able to meet most of their dietary needs through forage available in the
forage openings which was exhibited in the consistency of their diet among seasons. We
conclude that with regular maintenance forage openings can provide substantial forage
for elk throughout the year.
During the summer of 2012, a severe drought occurred across most of the
Midwest including Missouri. Drought has substantial effects on forage availability and
quality and herbivore diets (Frank and McNaughton 1982, Holecheck and Vavra 1983,
Stephenson et al. 1985, Van Horne et al. 1998). Although elk diets were consistent
throughout most seasons, we observed some changes in the elk diet during the 2012
drought in Missouri. Elk selected cool-season grasses during every season except
summer and fall 2012. Most plant species consumed in this group (e.g. orchardgrass) are
not native to our study area and are likely not resistant to drought. Grasses also begin to
70
senesce earlier under drought conditions and therefore decrease in digestibility (Van
Horne et al. 1998). Weedy forbs, woodland forbs, and other forbs were selected by elk in
2012, but not in 2011. These forage classes contain more native species and were likely
more resilient during the drought. Holecheck and Vavra (1983) found that forbs and
shrubs will provide more palatable forage than grasses during drought years. Similarly,
Van Horne et al. (1998) observed that under drought conditions native shrub habitats
provided a more stable food resource for Townsend’s ground squirrels (Spermophilus
townsendii) than perennial grasses and exotic annual forbs in grassland habitats in Idaho.
Although forage openings will provide large amounts of high quality forage within a
small area, management for native species may also be important to provide a more
reliable food source.
The reliability of correction factors has been questioned by some (Gill et al. 1983)
and supported by others (Pulliam and Nelson 1979, Vavra and Holecheck 1980, Leslie et
al. 1983, Bartolomé et al. 1995). Many wildlife studies using fecal samples to determine
diet composition choose to ignore the effects of differential digestibility of plant species
(McCullough 1985, Schneider 2006, Lupardus et al. 2011). However, the digestibility of
plants varies among species, seasons, and geographical regions (Pulliam and Nelson
1979). Including the effects of differential digestibility through correction factors can
have substantial effects on estimates of diet composition. For example, deciduous trees
decreased from 10% to 2.6% and legumes increased from 42% to 63% in the summer
diets of elk with the application of correction factors. According to our observations, we
believe, that when calculated carefully, correction factors will lead to more robust
estimates of ungulate diet composition from microhistological analysis of feces.
71
MANAGEMENT IMPLICATIONS
Elk managers in eastern states can minimize elk and human conflict by providing
enough forage on public lands to sustain elk populations. In densely forested areas,
creation of open lands may be necessary to provide forage for elk. Glades and open
woodlands within our study area do provide openings in the forest canopy for growth of
grasses and forbs. However, our results show that elk selected forage from forage
openings over forage available in natural open lands. The Missouri elk population is
relatively small (≈ 115 animals; unpublished data) and appears to currently have an
adequate food supply within forage openings on public lands. As the population grows
and requires more resources, the elk will either begin to utilize natural open lands more or
they will begin to search for food in privately owned pastures. Elk managers must
consider the costs and benefits of creating and maintaining forage openings both
ecologically and financially to determine how and where forage openings should be
implemented. In areas with large elk populations, it may not be feasible to incorporate
forage opening management (planting and maintenance) at a scale that will result in
substantial effects on elk reproduction and survival compared to the resources invested
(Cook 2002). However, eastern states that are seeking to maintain a small population of
elk may be able to implement forage opening management successfully at a reasonable
cost. We recommend scattering complexes of forage openings across the landscape to
minimize the impact on any single area. Within forage openings, we recommend
planting a mixture of cool-season grasses and grains to provide forage in the fall and
72
winter and clovers, alfalfa, and common lespedeza to provide forage during spring and
summer. We further recommend regular maintenance such as application of fertilizer
and lime, mowing to encourage fresh growth during summer, and replanting every 3-5
years as cool-season grasses tend to outcompete other species. Cost-share programs with
private landowners may be a worthwhile strategy to add more forage openings at a lower
cost. Leases with landowners to harvest hay from forage openings may also help offset
costs.
Management for native species is necessary to provide a stable resource for elk
particularly during periods when food is scarce. Managers should maintain natural open
lands through timber harvest and prescribed fire to promote growth of native grass and
forb species which may be an important food source during drought periods.
Regeneration after timber harvests will provide low-growing browse for elk.
Management of oak-hickory forests and woodlands for hard mast production will also
provide forage during fall and winter.
73
LITERATURE CITED
Alipayo, D., R. Valdez, J. L. Holechek, and M. Cardenas. 1992. Evaluation of
microhistological analysis for determining ruminant diet botanical composition.
Journal of Range Management 45:148-152.
Bartolomé, J., J. Franch, M. Gutman, and N. G. Seligman. 1995. Physical factors that
influence fecal analysis estimates of herbivore diets. Journal of Range
Management 48:267-270.
Chen, J., C. D. Huebner, S. R. Saunders, and B. Song. 2002. Plant distribution and
diversity across an Ozark landscape. Pages 45-65 in S. R. Shifley and J. M.
Kabrick, editors. Proceedings of the Second Missouri Ozark Forest Ecosystem
Project Symposium: Post-treatment Results of the Landscape Experiment.
General Technical Report NC-227. U. S. Department of Agriculture, Forest
Service, North Central Research Station, St. Paul, Minnesota, USA.
Cook, J. G. 2002. Nutrition and food. Pages 258-349 in D. E. Toweill, and J. W. Thomas,
editors. North American elk: ecology and management. Smithsonian Institution
Press, Washington, D. C., USA.
Daubenmire, R. F. 1959. A canopy-coverage method of vegetation analysis. Northwest
Science 33: 43-64.
Frank, D.A., and S. J. McNaughton. 1992. The ecology of plants, large mammalian
herbivores, and drought in Yellowstone National Park. Ecology 73:2043–2058.
Gill, R. B., L. H. Carpenter, R. M. Bartmann, D. L. Baker, and G. G. Schoonveld. 1983.
Fecal analysis to estimate mule deer diets. Journal of Wildlife Management
47:902-915.
74
Grabner, J. K. 2000. Ground layer vegetation in the Missouri Ozark Forest Ecosystem
Project: pre-treatment species composition, richness, and diversity. Pages 107-123
in S. R. Shifley and B. L. Brookshire, editors. Missouri Ozark Forest Ecosystem
Project: Site History, Soils, Landforms, Woody and Herbaceous Vegetation,
Down Wood, and Inventory Methods for the Landscape Experiment. General
Technical Report NC-208. U. S. Department of Agriculture, Forest Service, North
Central Research Station, St. Paul, Minnesota, USA.
Higgins, K. F., K. J. Jenkins, G. K. Clambey, D. W. Uresk, D. E. Naugle, R. W. Klaver,
J. E. Norland, K. C. Jensen, and W. T. Barker. 2012. Vegetation sampling and
measurement. Pages 381-409 in N. J. Silvy, editor. The Wildlife Techniques
Manual: Volume 1: Research. The John Hopkins University Press, Baltimore,
Maryland, USA.
Holecheck, J. L., and M. Vavra. 1983. Drought effects on diet and weight gain of
yearling heifers in northeastern Oregon. Journal of Range Management 36:227-
231.
Jenkins, K. J., and E. E. Starkey. 1991. Food habits of Roosevelt elk. Rangelands 13:261-
265.
Jenks, J. A., D. M. Leslie, Jr., R. L. Lochmiller, M. A. Melchiorsand, and W. D. Warde.
1989. Effects of compositing samples on analysis of fecal nitrogen. Journal of
Wildlife Management 53:213-215.
Kabrick, J. M., R. G. Jensen, S. R. Shifley, and D. R. Larsen. 2002. Woody vegetation
following even-aged, uneven-aged, and no-harvest treatments on the Missouri
Ozark Forest Ecosystem Project sites. Pages 84-101 in S. R. Shifley and J. M.
75
Kabrick, editors. Proceedings of the Second Missouri Ozark Forest Ecosystem
Project Symposium: Post-treatment Results of the Landscape Experiment.
General Technical Report NC-227. U. S. Department of Agriculture, Forest
Service, North Central Research Station, St. Paul, Minnesota, USA.
Keller, B. J. 2011. Factors affecting spatial and temporal dynamics of an ungulate
assemblage in the Black Hills, South Dakota. Dissertation, University of
Missouri, Columbia, Missouri, USA.
Kufeld, R. C. 1973. Foods eaten by the Rocky Mountain elk. Journal of Range
Management 26:106-113.
Leban, F. 1999. Resource Selection for Windows. Version 1.00 (Beta 8.4 – May 28,
1999).
Leslie, D. M., Jr., M. Vavra, E. E. Starkey, and R. C. Slater. 1983. Correcting for
differential digestibility in microhistological analyses involving common coastal
forages of the Pacific Northwest. Journal of Range Management 36:730-732.
Lupardus, J. L., L. I. Muller, and J. L. Kindall. 2011. Seasonal forage availability and diet
for reintroduced elk in the Cumberland Mountains, Tennessee. Southeastern
Naturalist 10:53-74.
Maehr, D. S., R. Grimes, and J. L. Larkin. 1999. Initiating elk restoration in the east: the
Kentucky case study. Proceedings of the Annual Conference of Southeastern Fish
and Wildlife Agencies 53:350363.
Manly, B. F. J., L. L. McDonald, D. Thomas, T. L. McDonald, and W. P. Erickson. 2002.
Resource selection by animals: statistical design and analysis for field studies.
Second edition. Kluwer Academic, Boston, Massachusetts, USA.
76
McCullough, D. R. 1985. Variables influencing food habits of white-tailed deer on the
George Reserve. Journal of Mammalogy 66:682-692.
McShea, W. J., and G. Schwede. 1993. Variable acorn crops: responses of white-tailed
deer and other mast consumers. Journal of Mammalogy 74:999-1006.
Meinert, D., T. Nigh, and J. Kabrick. 1997. Landforms, geology, and soils of the MOFEP
study area. Pages 56-68 in S. R. Shifley and B. L. Brookshire, editors.
Proceedings of the Missouri Ozark Forest Ecosystem Project Symposium: An
Experimental Approach to Landscape Research. General Technical Report NC-
193. U. S. Department of Agriculture, Forest Service, North Central Research
Station, St. Paul, Minnesota, USA.
Mendiburu, F. (2014). agricolae: Statistical Procedures for Agricultural Research. R
package version 1.2-1. http://CRAN.R-project.org/package=agricolae. Accessed
29 Dec 2014.
Missouri Department of Conservation (MDC). 2010. Elk Restoration in Missouri.
Missouri Department of Conservation, Jefferson City, USA.
Missouri Department of Conservation [MDC]. 2013. Peck Ranch Conservation Area.
<http://mdc4.mdc.mo.gov/applications/moatlas/AreaSummaryPage.aspx?txtAreaI
D=5203>. Accessed 20 Mar 2013.
Murrow, J. L. 2007. An experimental release of elk into Great Smoky Mountains
National Park. Thesis, University of Tennessee, Knoxville, Tennessee, USA.
Norbury, G. L., and G. D. Sanson. 1992. Problems with measuring diet selection of
terrestrial, mammalian herbivores. Australian Journal of Zoology 17:1-7.
77
Pulliam, D. E., Jr., and J. R. Nelson. 1979. Determination of digestibility coefficients for
quantification of fecal analysis with elk. Pages 240-246 in M. S. Boyce and L. D.
Hayden-Wing, eds. North American elk: ecology, behavior, and management.
Univ. Wyoming, Laramie.
Rocky Mountain Elk Foundation (RMEF) and Southern & Eastern Kentucky Tourism
Development Association. 2007. Study of the elk and wildlife viewing potential
for southern and eastern Kentucky. 163 pp.
Schneider, J., D. S. Maehr, K. J. Alexy, J. J. Cox, J. L. Larkin, and B. C. Reeder. 2006.
Food habits of reintroduced elk in southeastern Kentucky. Southeastern Naturalist
5:535-546.
Skolvin, J. M., P. Zager, and B. K. Johnson. 2002. Elk habitat selection and evaluation.
Pages 530-555 in D. E. Toweill, and J. W. Thomas, editors. North American elk:
ecology and management. Smithsonian Institution Press, Washington, D. C.,
USA.
Stephenson, T. E., J. L. Holecheck, and C. B. Kuykendall. 1985. Drought effect on
pronghorn and other ungulate diets. Journal of Wildlife Management 49:146-151.
Tanentzap, A. J., J. N. Bee, W. G. Lee, R. B. Lavers, J. A. Mills, A. F. Mark, and D. A.
Coomes. 2009. The reliability of palatability estimates obtained from rumen
contents analysis and a field-based index of diet selection. Journal of Zoology
278:243–248.
Vangilder, L. D. 1997. Acorn production on the Missouri Ozark Forest Ecosystem
Project sites: pre-treatment data. Pages 198-209 in S. R. Shifley and B. L.
Brookshire, editors. Proceedings of the Missouri Ozark Forest Ecosystem Project
78
Symposium: An Experimental Approach to Landscape Research. General
Technical Report NC-193. U. S. Department of Agriculture, Forest Service, North
Central Research Station, St. Paul, Minnesota, USA.
Van Horne, B., R. L. Schooley, and P. B. Sharpe. 1998. Influence of habitat, sex, age,
and drought on the diet of Townsend’s ground squirrels. Journal of Mammalogy
79:521-537.
Vavra, M., and J. L. Holechek. 1980. Factors influencing microhistological analysis of
herbivore diets. Journal of Range Management 33:371-374.
Waller, R. A., and D. B. Duncan. 1969. A Bayes rule for the symmetric multiple
comparisons problem. Journal of the American Statistical Association 64:1484-
1503.
Wildlife Habitat and Nutrition Lab, Department of Natural Resources Sciences,
Washington State University, Pullman, Washington, USA.
Xu, M., S. C. Saunders, and J. Chen. 1997. Analysis of landscape structure in the
southeastern Missouri Ozarks. Pages 41-55 in S. R. Shifley and B. L. Brookshire,
editors. Proceedings of the Missouri Ozark Forest Ecosystem Project Symposium:
An Experimental Approach to Landscape Research. General Technical Report
NC-193. U. S. Department of Agriculture, Forest Service, North Central Research
Station, St. Paul, Minnesota, USA.
79
TABLES
Table 1. Seasonal diet composition (%) before and after digestion trials and correction factors (CF) for elk in Missouri.
aCorrection factors for the “other” forage class were calculated assuming no change after digestion. Forage classes with correction
factors greater than this class were overestimated in the original diet for that season. Classes with correction factors less than the
“other” class were underestimated in the original diet.
b Pine was the standard plant added to each diet and was only used to help calculate the rest of the correction factor.
Summer
Fall
Winter
Spring
Forage class Before After CF
Before After CF
Before After CF
Before After CF
Othera
50.00 50.00 1.87
50.00 50.00 1.86
50.00 50.00 1.61
50.00 50.00 1.73
Acorn
19.60 15.33 2.38
Browse 13.41 31.33 0.80
18.92 21.33 1.65
22.12 32.67 1.09
13.74 12.67 1.88
Forbs 43.65 18.00 4.53
27.84 30.00 1.73
29.63 21.33 2.23
59.29 28.00 3.67
Grasses 42.94 41.33 1.94
33.64 24.00 2.61
48.25 38.00 2.04
26.96 50.67 0.92
Pineb
5.00 9.33 1.00
5.00 9.33 1.00
4.97 8.00 1.00
5.00 8.67 1.00
80
Table 2. Seasonal diet selection ranks from summer 2011 - winter 2012 for elk in Missouri.
Summer
Fall
Winter
Spring
Forage class 2011 2012
2011 2012
2011 2012
2012
Cool-season grasses 2a (B
b) 2 (B)
1.5 (A) 5 (BC)
2 (B) 2 (B)
1 (A)
Deciduous trees 12 (F) 11.5 (F)
9 (CD) 10 (DE)
9 (F) 9 (DE)
7 (C)
Evergreen trees 6 (CD) 7 (DE)
4 (B) 8 (BCD)
7 (DE) 6 (CD)
3 (B)
Grains 1 (A) 1 (A)
1.5 (A) 1 (A)
1 (A) 1 (A)
2 (A)
Legumes 4 (C) 6 (CD)
7 (B) 6.5 (BCD)
3 (C) 3 (B)
5 (B)
Other forages 9 (E) 10 (F)
10 (CD) 12 (E)
10 (FG) 8 (CDE)
10 (CDE)
Other forbs 7 (DE) 5 (CD)
12 (D) 6.5 (BCD)
11 (G) 10 (DE)
9 (CDE)
Sedges 10 (E) 8 (E)
5 (B) 4 (B)
4 (C) 5 (C)
4 (B)
Shrubs and vines 11 (F) 11.5 (F)
11 (CD) 11 (DE)
12 (H) 12 (F)
12 (E)
Warm-season grasses 5 (CD) 9 (E)
3 (B) 9 (CDE)
8 (E) 11 (EF)
11 (DE)
Weedy forbs 3 (C) 3 (BC)
6 (B) 2.5 (B)
6 (D) 7 (CDE)
6 (B)
Woodland forbs 8 (E) 4 (BCD)
8 (C) 2.4 (B)
5 (C) 4 (B)
8 (CD)
a selection ranks are from most selected (1) to least selected (12).
b ranks with the same letter were not significantly (P > 0.05) different from each other.
81
Table 3. Seasonal diet composition (%) by forage class for elk in Missouri, 2011-2012.
Summer
Fall
Winter
Spring
Forage Class 2011 2012
2011 2012
2011 2012
2012
Overall
Cool-season grasses 7.5 2.3
15.5 1.8
16.5 12.8
4.3
8.6
Grains 9.4 8.2
6.2 8.5
16.9 12.7
3.3
9.3
Hard mast 0.1 0.1
21.2 3.7
7.5 0.7
2.3
5.0
Legumes 55.5 64.0
19.5 36.4
29.7 32.4
74.8
45.5
Other Forage 1.6 0.0
0.7 0.3
2.1 7.1
0.2
1.7
Other forbs 3.9 7.4
0.4 5.9
0.0 2.1
0.3
3.0
Carex spp. 0.0 0.3
1.7 3.3
6.8 5.2
2.0
2.7
Shrubs & Vines 0.9 0.8
2.4 4.4
0.3 3.3
1.5
1.9
Deciduous Trees 2.7 3.0
10.5 12.3
10.5 10.7
8.6
8.1
Evergreen Trees 0.0 0.0
0.0 0.0
0.3 0.5
0.3
0.2
Unknown (not ID) 4.0 4.7
5.3 5.9
3.9 5.1
1.4
4.3
Weedy Forbs 1.1 2.2
0.6 4.1
0.1 0.8
0.3
1.3
Woodland Forbs 2.1 3.7
0.3 7.8
0.2 2.4
0.2
2.3
Warm-season Grasses 11.2 3.3 15.6 5.4 5.2 4.3 0.6 6.3
82
Table 4. Seasonal diet composition (%) of cultivated forage plants for elk in Missouri summer 2011-winter 2012.
Summer Fall Winter Spring Overall
Species Common Name 2011 2012 2011 2012 2011 2012 2012 Average
Andropogon spp. Bluestem 3.6 0.1
4.3 0.5
0.5 0.1
1.2
Dactylis glomerata Orchardgrass 0.5 0.1
7.9
9.1 4.5
0.8
3.3
Hordeum vulgare Barley
0.2
7.7
4.2
1.6
Medicago sativa Alfalfa 12.1 15.1
3.8 8.3
9.2 5.9
19.0
10.7
Secale cereale Rye 5.8 1.3
2.2 0.1
3.4 1.0
0.9
2.0
Sorghastrum nutans Indiangrass 0.2 0.2
1.7
0.1
0.3
Trifolium spp. Clover 10.8 23.0
4.1 20.4
17.7 16.6
45.7
20.1
Tripsacum
dactyloides Gammagrass
0.4
0.1
Triticum aestivum Wheat 3.7 5.5
4.0
13.5 6.6
2.5
5.3
Total 36.8 45.5 28.4 37.0 53.5 39.0 68.9 44.6
83
FIGURES
Figure 1. Missouri elk restoration zone map depicting land ownership. The Missouri
Department of Conservation and Missouri Spatial Data Information Service (MSDIS)
were the sources for these GIS layers.
84
Figure 2. Map of Peck Ranch Conservation Area. These GIS layers were provided by the
Missouri Department of Conservation, Missouri Spatial Data Information Service
(MSDIS), and the U.S. Census Bureau.
85
Figure 3. Representation of placement of quadrats along transects at vegetation sampling
points. The 30m transect is run in a randomly selected cardinal direction starting from the
sampling point (large dot) with 0.1m2 quadrats placed every meter. This figure only
represents ½ (15m) of a transect.
1m
86
87
APPENDIX I. SEASONAL DIET COMPOSITION AND AVAILABILITY OF
PLANT SPECIES OBSERVED IN THE MISSOURI OZARKS.
Forage Class Species Summer Fall Winter Spring
Cool-season grassesa
4.3 9.3 14.7 4.3
CSG Arundinaria gigantea 0.1 xb x x
CSG Bromus spp. 0.4 1.4 1.8 0.1
CSG Cynodon dactylon x x
x
CSG Dactylis glomerata 0.3 4.3 6.8 0.8
CSG Danthonia spicata x
x
CSG Elymus spp. 2.9 1.5 1.8 2.7
CSG Festuca spp. x x 0.1 x
CSG Lolium multiflorum x x x x
CSG Phleum pratense x
x
CSG Poa spp. 0.7 1.9 4.1 0.6
CSG Sporobolus spp. x 0.3 0.0 x
CSG Vulpia octoflora x
x
Deciduous trees
2.9 11.3 10.6 8.6
D_Trees Acer rubrum 0.4 1.0 0.2 x
D_Trees Amelanchier x 0.1 x x
D_Trees Asimina parviflora x x x x
88
Forage Class Species Summer Fall Winter Spring
D_Trees Carpinus caroliniana x x x x
D_Trees Carya spp. 0.2 0.3 0.5 0.5
D_Trees Celtis occidentalis x x x x
D_Trees Celtis tenuifolia x x x x
D_Trees Cercis canadensis x 0.4 0.0 x
D_Trees Cornus spp. 1.0 5.3 1.3 1.6
D_Trees Diospyros virginiana x x x x
D_Trees Fraxinus americana x x x x
D_Trees Juglans nigra x x x x
D_Trees Morus rubra x x x x
D_Trees Nyssa sylvatica x x x x
D_Trees Platanus occidentalis x x x x
D_Trees Prunus serotina 0.1 0.1 x x
D_Trees Quercus spp. 1.3 3.3 8.5 6.4
D_Trees Rhamnus caroliniana 0.0 0.6 0.1 x
D_Trees Sassafras albidum 0.0 0.3 0.0 x
D_Trees Sideroxylon lanuginosum x x x x
D_Trees Tilia americana x x x x
D_Trees Ulmus spp. x x x x
Evergreen Trees
0.0 0.0 0.4 0.3
89
Forage Class Species Summer Fall Winter Spring
E_Trees Juniperus virginiana x x 0.0 x
E_Trees Pinus echinata 0.0 x 0.4 0.3
Grains
8.7 7.3 14.8 3.3
Grains Hordeum vulgare 0.1 3.5 2.1 x
Grains Millet 0.4 0.3
Grains Secale cereale 3.0 1.3 2.2 0.9
Grains Triticum aestivum 4.8 2.2 10.1 2.5
Grains Zea mays 0.3
0.4 x
Hard mast
0.1 13.3 4.1 2.3
Legumes
60.7 27.2 31.0 74.8
Legumes Amphicarpaea bracteata x x
Legumes Baptisia bracteata x
x
Legumes Cassia obtusifolius x
x x
Legumes Chamaecrista fasciculata x x
x
Legumes Chamaecrista nictitans x x
x
Legumes Coronilla varia x x x x
Legumes Dalea purpurea 2.0 0.2
5.5
Legumes Desmodium spp. 2.4 0.4
x
Legumes Galactia regularis x x
x
Legumes Kummerowia spp. 16.0 6.1 5.5 4.2
90
Forage Class Species Summer Fall Winter Spring
Legumes Lespedeza cuneata 3.9 2.0 0.1 0.0
Legumes Lespedeza spp. 3.9 1.1 0.7 0.3
Legumes Lotus corniculatus x
Legumes Medicago lupulina x x
x
Legumes Medicago sativa 14.0 5.8 7.5 19.0
Legumes Mimosa quadrivalvis x x
x
Legumes Senna marilandica x x
x
Legumes Stylosanthes biflora x x
x
Legumes Tephrosia virginiana x x
x
Legumes Trifolium spp. 18.3 11.5 17.2 45.7
Legumes Vicia spp. 0.3 0.1 0.0 x
Other forage
0.6 0.5 4.6 0.2
Other forage Fern 0.1 0.5 4.6 0.2
Other forage Lichen 0.5 x x x
Other forage Moss x x x x
Other forbs (both disturbed and natural habitats) 6.1 2.9 1.0 0.3
Other forbs Apocynum cannabinum x x
x
Other forbs Asclepias spp. x x
x
Other forbs Boechera spp. x x
x
Other forbs Cirsium spp. 0.4 0.3
x
91
Forage Class Species Summer Fall Winter Spring
Other forbs Comandra umbellata x x
x
Other forbs Coreopsis lanceolata x x
x
Other forbs Coreopsis palmata x x
x
Other forbs Croton willdenowii x x
Other forbs Cynanchum laeve x x
x
Other forbs Elephantopus carolinianus x x
x
Other forbs Eupatorium spp. 0.0 x
x
Other forbs Euphorbia corollata x x
x
Other forbs Euphorbia dentata x x
x
Other forbs Euphorbia maculata x x
x
Other forbs Fragaria virginiana 0.0 x
x
Other forbs Galium spp. 2.2 0.5 x 0.1
Other forbs Geranium spp. 0.2 x x x
Other forbs Glandularia canadensis x x
x
Other forbs Hedyotis spp. x
x
Other forbs Helianthus hirsutus 0.7 0.4 0.2 x
Other forbs Hieracium gronovii x x
x
Other forbs Impatiens capensis 0.7 0.5
Other forbs Ipomoea spp. x x
x
Other forbs Krigia biflora x x
x
92
Forage Class Species Summer Fall Winter Spring
Other forbs Lactuca spp. 0.9 x
x
Other forbs Liatris spp. x x
x
Other forbs Lobelia inflata x x
x
Other forbs Lysimachia lanceolata x x
x
Other forbs Oxalis spp. x x x x
Other forbs Parthenium integrifolium x x
x
Other forbs Phlox pilosa x x
x
Other forbs Physalis virginiana x x
x
Other forbs Polygonum spp. x x 0.3 x
Other forbs Potentilla simplex x x x x
Other forbs Ratibida pinnata x x
x
Other forbs Rudbeckia spp. 0.1 0.9 0.2 x
Other forbs Salvia lyrata x x
x
Other forbs Silene virginica x x
x
Other forbs Solidago spp. 0.7 0.3 0.4 x
Other forbs Stachys pilosa x x
0.2
Other forbs Verbena urticifolia x x
x
Other forbs Verbesina spp. x x
x
Other forbs Vernonia spp. x x
x
Other forbs Viola spp. 0.2 0.0
x
93
Forage Class Species Summer Fall Winter Spring
Sedges
0.2 2.5 6.0 2.0
Sedges Carex spp. 0.2 2.5 6.0 2.0
Sedges Eleocharis compressa x x
x
Shrubs and vines
0.8 3.3 1.8 1.5
Shrubs/vines Berchemia scandens x x x x
Shrubs/vines Campsis radicans x x x x
Shrubs/vines Ceanothus americanus 0.0 x 1.0 x
Shrubs/vines Corylus americanus x x x x
Shrubs/vines Crataegus spp. x x x x
Shrubs/vines Dirca palustris x x x x
Shrubs/vines Elaeagnus umbellata 0.6 1.0 0.3 1.5
Shrubs/vines Hamamelis spp. x x x x
Shrubs/vines Hydrangea arborescens x x x x
Shrubs/vines Hypericum spp. x x x x
Shrubs/vines Lindera benzoin x x x x
Shrubs/vines Lonicera flava x x x x
Shrubs/vines Lonicera japonica x x x x
Shrubs/vines Lonicera maackii x x x x
Shrubs/vines Menispermum canadense x x x x
Shrubs/vines Ostrya virginiana x x x x
94
Forage Class Species Summer Fall Winter Spring
Shrubs/vines Parthenocissus quinquefolia 0.1 0.2 x x
Shrubs/vines Rhus spp. 0.0 0.3 x x
Shrubs/vines Ribes missourense x x x x
Shrubs/vines Rosa spp. x x x x
Shrubs/vines Rubus spp. x x x x
Shrubs/vines Smilax spp. 0.1 0.1 0.0 x
Shrubs/vines Symphoricarpos orbiculatus x x x x
Shrubs/vines Toxicodendron radicans x x x x
Shrubs/vines Vaccinium spp. x 0.6 0.1 x
Shrubs/vines Vibrnum rufidulum x x x x
Shrubs/vines Vitis spp. 0.1 1.2 0.3 x
Unknown (not identified) 4.4 5.5 4.5 1.4
Weedy forbs (disturbed habitats) 1.8 2.2 0.5 0.3
Weedy forbs Achillea millefolium 0.1 x x x
Weedy forbs Allium spp. x x x x
Weedy forbs Ambrosia artemisiifolia 0.1 0.1
x
Weedy forbs Ambrosia bidentata x x
Weedy forbs Anaphalis spp. x 1.4
Weedy forbs Arabidopsis thaliana x
x x
Weedy forbs Barbarea vulgaris x x x x
95
Forage Class Species Summer Fall Winter Spring
Weedy forbs Bidens spp. 0.5 0.1
x
Weedy forbs Brassica spp. 0.1 0.1 x x
Weedy forbs Capsella bursa-pastoris x x x x
Weedy forbs Cerastium vulgatum x x x x
Weedy forbs Chenopodium spp. x
x
Weedy forbs Cichorium intybus x
x
Weedy forbs Convovuluus spp. x x 0.1 x
Weedy forbs Conyza canadensis 0.7 0.3
x
Weedy forbs Croton capitatus x x
Weedy forbs Croton glandulosus x x
Weedy forbs Daucus carota x x x x
Weedy forbs Diodia teres x x
x
Weedy forbs Erechtites hieracifolia x x
x
Weedy forbs Erigeron spp. 0.1 0.1 0.1 x
Weedy forbs Erodium cicutarium x x x x
Weedy forbs Helenium amarum x x
Weedy forbs Lamium amplexicaule x
x
Weedy forbs Lepidium virginicum x
x x
Weedy forbs Marrubium vulgare x
x
Weedy forbs Mazus japonicus x
x x
96
Forage Class Species Summer Fall Winter Spring
Weedy forbs Mollugo verticillata x
x
Weedy forbs Perilla frutescens x x
x
Weedy forbs Phytolacca americana x x
x
Weedy forbs Plantago spp. 0.3 0.1 0.3 0.3
Weedy forbs Portulaca oleracea x
x
Weedy forbs
Pseudognaphalium
obtusifolium x x
x
Weedy forbs Pyrrhopappus carolinianus x x x x
Weedy forbs Rumex crispus x x x x
Weedy forbs Rumex obtusifolius x x x x
Weedy forbs Solanum carolinense x x
x
Weedy forbs Solanum ptycanthum x x
x
Weedy forbs Stellaria media x x
x
Weedy forbs Taraxacum spp. x x x x
Weedy forbs Thlaspi arvense x x x x
Weedy forbs Verbascum thapsus x x x x
Weedy forbs Veronica arvensis x
x x
Woodland and glade forbs (natural habitats) 3.1 3.7 1.3 0.2
Woodland forbs Actaea racemosa x x
x
Woodland forbs Ageratina altissima x x
x
97
Forage Class Species Summer Fall Winter Spring
Woodland forbs Agrimonia ssp. x x
x
Woodland forbs Anemone spp. x x
x
Woodland forbs Antennaria plantaginifolia x x
x
Woodland forbs Aristolochia serpentaria x
x
Woodland forbs Aureolaria spp. x x
x
Woodland forbs Campanula americana 0.1 x 0.1 x
Woodland forbs Circaea lutetiana x x
x
Woodland forbs Clinopodium arkansanum x x
x
Woodland forbs Cunila organoides x x
x
Woodland forbs Dioscorea spp. x x
x
Woodland forbs Echinacea simulata 2.0 0.9 0.4 0.2
Woodland forbs Gillenia stipulata x x
x
Woodland forbs Heliopsis helianthoides x x
x
Woodland forbs Matelea decipiens x x
x
Woodland forbs Mianthemum racemosum 0.7 0.1 0.2 x
Woodland forbs Monarda bradburiana 0.1 0.5 0.1 x
Woodland forbs Oenothera linifolia x
x
Woodland forbs Passiflora lutea x x x x
Woodland forbs Penstemon pallidus x 1.4 0.5 x
Woodland forbs Persicaria virginiana x x
x
98
Forage Class Species Summer Fall Winter Spring
Woodland forbs Phlox divaricata x x
x
Woodland forbs Phyrma leptostachya x x
x
Woodland forbs Potentilla canadensis x x
x
Woodland forbs Prenanthes altissima x x
x
Woodland forbs Pycananthemum tenuifolium x x
x
Woodland forbs Ranunculus hispidus x x
x
Woodland forbs Ruellia spp. x x
x
Woodland forbs Sanicula canadensis x x
x
Woodland forbs Sanicula odorata x x
x
Woodland forbs Scutellaria elliptica x x
x
Woodland forbs Scutellaria incana x x
x
Woodland forbs Silene stellata x x
x
Woodland forbs Silphium asteriscus x x
x
Woodland forbs Silphium terebinthinaceum x x
x
Woodland forbs Symphyotrichum (Aster) spp. 0.2 0.7
x
Woodland forbs Talinum calycinum x x
x
Woodland forbs Teucrium canadense x x
x
Woodland forbs Thalictrum thalictroides x
x
Woodland forbs Thaspium trifoliatum x x
x
Woodland forbs Tradescantia spp. x
x
99
Forage Class Species Summer Fall Winter Spring
Woodland forbs Uvularia grandiflora x x
x
Woodland forbs Zizia aurea x x
x
Warm-season grasses 6.3 11.0 4.8 0.6
WSG Andropogon spp. 1.4 2.6 0.3 x
WSG Bouteloua curtipendula 0.4 0.2 0.1 x
WSG Brachyelytrum erectum x x
x
WSG Chasmanthium latifolium x x x x
WSG Dactyloctenium aegyptium x
WSG Digitaria ischaemum x x
x
WSG Erianthus alopecuroides x x x x
WSG Leersia virginica x x
x
WSG Muhlenbergia spp. 0.0 x
x
WSG
Panicum spp./Dicanthileum
spp. 2.7 4.0 3.6 0.5
WSG Schizachyrium scoparium 0.3 x x x
WSG Setaria spp. 1.2 3.1 0.7 0.1
WSG Sorghastrum nutans 0.2 0.9 0.0 x
WSG Tridens flavus x x
x
WSG Tripsacum dactyloides x 0.2 x x
100
a Italicized lines indicate the name of forage class and the total diet composition for that
forage class
b “X” indicates that the plant species was available, but was not observed in the diet
during that season. Blank spaces indicate that the plant species was not available during
that season.