Development of habitat suitability for the black-tailed prairie dog (cynomys ludovicianus) in Colorado

Material Information

Development of habitat suitability for the black-tailed prairie dog (cynomys ludovicianus) in Colorado an application of geospatial modeling
Mit, Adel
Publication Date:
Physical Description:
xi, 61 leaves : ; 28 cm


Subjects / Keywords:
Black-tailed prairie dog -- Colorado ( lcsh )
Habitat suitability index models -- Colorado ( lcsh )
Black-tailed prairie dog ( fast )
Habitat suitability index models ( fast )
Colorado ( fast )
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )


Includes bibliographical references (leaves 54-61).
General Note:
Department of Geography and Environmental Sciences
Statement of Responsibility:
Adel Mit.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
318534811 ( OCLC )
LD1193.L547 2008m M57 ( lcc )

Full Text
Adel Mit
B.S., Kazakh State National University, Almaty, Kazakhstan, 1996
A thesis submitted to the
University of Colorado Denver
in partial fulfillment
of the requirements for the degree of
Master of Science
Environmental Sciences

This thesis for the Master of Science
degree by
Adel Mit
has been approved
John Wyckcff
Jon Barbour

Mit, Adel (M.S., Environment Sciences)
Development of Habitat Suitability Indices for the Black-Tailed Prairie Dog (Cynomys
Ludovicianus) in Colorado: an Application of Geospatial Modeling
Thesis directed by Professor John Wyckoff
The Habitat Suitability Indices model for the black-tailed prairie dog was developed
using the geospatial modeling capabilities of ArcGIS9.2 Software and multi-criteria
evaluation models. The modeling procedure was based on habitat variables, which cover
the most essential habitat characteristics of species preferred habitats as described in the
peer-reviewed literature and including personal contacts with species experts in
Colorado, USA. In the models, key variables that apply to respective species are analyzed
and results are displayed on a computer monitor or printed in hardcopy format. Habitat
for a species is rated from zero (unsuitable) to five (highly suitable) on the map.
Variables used in the models include land cover class, elevation limits, slope, and
hydrology data. It is relatively easy to modify the models, e.g. to include different
viewpoints about variables that determine suitable habitat for the black-tailed prairie dog.
Where capture or sampling data are available, the results of the models are tested using
spatial statistical methods. GIS data of the black-tailed prairie dog colonies for the year
2000 were used to verify the ability to accurately predict the HSI model output between
the suitable habitat areas and breeding site locations of the species. Validation of the HSI
model provides evidence that the model works. The HSI model provides important

information for conservation biologists and land managers concerned with preserving
black-tailed prairie dog diversity in the Comanche National Grassland, Colorado, USA.
This abstract accurately represents the content of the candidates thesis. I recommend its

I dedicate this thesis to my parents and family for their unfaltering support and
understanding while I was completing this thesis.

My thanks to my advisor, Dr. John Wyckoff, for his contribution and support
though the research project. I also wish to thank Fred Nuszdorfer, for his
assistance and help throughout this research. My thanks to Dr.Lynn Johnson
and Dr. Jon Barbour for their assistance and serving on my thesis committee.

Figures........................................................... x
Tables............................................................ xi
Wildlife Habitat Suitability Models...................... 1
Basic Ecology and Status of Black-Tailed Prairie Dog..... 4
Habitat Requirements..................................... 7
Recent Research........................................... 10
Objectives................................................ 13
Study Area................................................ 15
Spatial Database Development.............................. 17
Habitat Suitability Mapping............................... 19
Weighted Overlay Technique............................... 20
Model Construction and Mapping........................... 24
Methods of Validation for HSI Model...................... 29
Sensitivity Analysis Approach............................ 31
Validation of the HIS Model.............................. 33

Sensitivity analysis............................ 40
Summary......................................... 45
Datasets of GIS coverages used for creation the HSI model of BTPD in the
Carrizo study area, the Comanche National Grassland..... 50
BIBLIOGRAPHY............................................ 53

2.1 Map of the Carrizo study area in the Comanche National Grassland,
Colorado........................................................ 16
2.2 Approach of solving spatial problems for the HSI model for BTPD... 20
2.3 Steps involved in HSI model for BTPD in Carrizo study area....... 21
3.1 Output map from HSI model for the BTPD........................... 36
3.2 Sensitivity analysis, where weighs 0.5, 0.7 and 1.0 were applied for
vegetation characteristics...................................... 43
3.3 The ModelBuilder flow diagram for the HSI model of the BTPD... 44

2.1 Slope variables and six habitat classes within the study area of 26
3.1 Habitat Suitable areas by Zone in the Carrizo unit............... 38
3.2 One-way ANOVA Test (Percent vs by Zone) for the HSI model for 39
3.3 Individual 95% CIs For Mean Based on Pooled StDev................ 39
3.4 One-Sample T: Zone 3. Test of Mean = 0.448 vs not=0.448.......... 39
3.5 Sensitivity analyses, where weights 0.5, 0.6, 07, 0.8, 0.9 and 1.0
were applied for vegetation characteristics..................... 41

Wildlife Habitat Suitability Models
Geographic Information Systems (GIS) are assemblages of computer
programs that perform a variety of manipulations of remote sensing, cartographic
digital data and tabular data (Bolstad 2005). GISs may include the capability to
combine several spatially registered maps in different ways, analyze the
neighborhood of each cell on a map, and measure distance within the map or
parcel size of different cover types. Most GISs have the capability to generate
statistics and produce tabular descriptions in addition to maps. Capabilities of
GISs make them powerful tools for modeling wildlife habitat, especially where
spatial variables are relevant (Bolstad 2005). The possibility of synchronous
consideration for many variables over large areas and the resulting generation of
data in digital cartographical formats represent important advantages of GIS. By
using sequences of GIS operations and key habitat variables, it is possible to build
models that will automatically generate habitat suitability maps (Store and Kangas
2001). Habitat and other important environmental key variables for example, land
cover, elevation, road network data, or any other variable that can be represented
cartographically according to the species habitat preference (Store and Kangas

2001). Developments of GIS habitat models are possible whenever the
environment data for controlling habitat suitability are available for database
development. The quality of habitat models depends on the accuracy of the
database, the selection of key variables, and the sequence of rules that mimic the
biology of the species. The participation of experts who have detailed in the
habitat suitability modeling process (Store and Jokimaki 2003; Malczewsk 2004).
Decades of studying wildlife-habitat relationships by many researches
have provided important insights into habitat requirements for many species. Data
from these studies have been used to develop wildlife habitat models such as
habitat suitability index (HSI) models. HSI models translate existing knowledge
of a species' habitat requirements into quantitative measures of habitat quality.
HSI models were first developed in the 1970s and 1980s to assist in impact
assessment, project planning, habitat management, and in understanding species
habitat relationships (USFWS 1981a). Many HSI models were developed using
the Delphi Process (Crance 1987). The Delphi process relies on extensive
knowledge of the species being studied. HSI models describe habitat suitability
for a particular species, not population sizes or trends. HSI models quantify
habitat suitability for species based on environmental variables thought to be
important to species presence. The development of HSI models is based on a
thorough analysis of the habitat requirements of a species and conclusion of

expert opinion (Morrison et al. 2006). HSI models evaluate heavily weighted
resource attributes important to species survival, or reproduction. These models
translate existing knowledge of a species' habitat requirements into quantitative
measures of habitat quality. They are useful in natural resource planning by
predicting the impacts of resource management practices on wildlife habitat. HSI
models, developed by USFWS, are based on data collected in the field for each
species modeled. These models are only valid where identical field data are
colleted for each area being studied (USFWS 1981a).
The Colorado Department of Transportation together with University of
Colorado, Denver (CDOT/UCD) developed HSI models that relied largely on
datasets derived from remotely sensed imagery and other publicly available data
acquired from various agencies (Wyckoff et al. 2004; Wyckoff et al. 2007). HSI is
a numerical index that represents the ability of a given habitat to provide life
requisites for a species on a relative scale with ranges from 0 to 1. The value of 0
is totally unsuitable, and value of 1 is totally suitable (USFWS 1981a). In these
HSIs, key environmental variables are weighted and scaled in ArcGIS9.2
software based on expert opinion. For instance, grass-dominated vegetation
identified from satellite imagery was given the highest weight in the model for
identifying suitable habitat for the Kit Fox (Wyckoff et al. 2007). The result of
these HSI models output is a map that depicts the location of suitable habitat

(Gustafson et al. 2001; Marzluff et al. 2002; Larson et al. 2004; Wyckoff et al.
2004; Wyckoffet al. 2007).
Basic ecology and status of the black-tailed prairie dog
Prairie dogs (Cynomys spp.) are an important species in North American
grasslands. The black-tailed prairie dog (BTPD) (Cynomys ludovicianus) is
distinguished from the other species of prairie dogs by their geographic map of
range. The BTPD is a large, colonial, burrowing rodent of the squirrel family
(Sciuridae) found on the short-and mixed-grass prairies of the Great Plains region.
The species occurs at elevations ranging from 915 to 1,830 meters. They dig
extensive burrow systems with large mounds 15-20 cm high (Proctor and Haskins
1998; Proctor et al. 2006). Prairie dogs have a profound impact on abiotic and
biotic features of their ecosystems. The species can influence environmental
heterogeneity, plant succession, hydrology, nutrient cycling and biodiversity. This
spices creates important interactions in prairie and steppe ecosystems that may
enhance total diversity (Miller et al. 1994; Delting 1998; Bangert and
Slobodchikoff 2000; Miller et al. 2000). Prairie dogs significantly reduce
vegetation height, species richness and cover, and excavate burrows that offer
shelter for other grassland species (Kotliar 2000). The associated localized

disturbance brought on by the presence of prairie dog colonies is thought to
beneficially affect grassland biodiversity at broader spatial scales (Delting 1998).
Plant species composition on BTPD towns generally shifts over time from
grasses to forbs in northern mixed grass and shortgrass prairies (Whicker and
Detling 1998; Winter et al. 2002). Grazing by BTPD leads to increases in bare
ground and lower canopy height (Delting 1998; Guenther and Detling 2003).
These alterations and their subsequent influences on associated species have led
to prairie dogs being characterized as a keystone species in short-and mixed-grass
prairie ecosystems of western North America (Miller et al. 1994; Bangert and
Slobodchikoff 2000; Kotliar 2000; Miller et al. 2000). Keystone species are
defined as those that have a disproportionate effect on the structure and function
of an ecosystem relative to their abundance, and performing a role unlike any
other species in the community (Coppock et al. 1983; Kotliar 2000). Prairie dogs
have been found to influence diversity of birds (Agnew et al. 1986), small
mammals (Ceballos et al. 1999), reptiles, amphibians (Shipley and Reading
2006), predators (Clark et al. 1982) and arthropods (Russell and Detling 2003).
Many species that rely on prairie dogs and their colonies could disappear if prairie
dogs were to go extinct, or even be locally extirpated (Miller et al. 1994).
The historic geographic range of the BTPD includes most of the Great
Plains Region of North America from eastern Nebrask, west to the foothills of the

Rocky Mountains, and from southern Saskatchewan in Canada south to northern
Mexico. Prairie dogs inhabited portions of: Montana, North Dakota, South
Dakota, Wyoming, Colorado, Nebraska, Kansas, Oklahoma, Texas, Arizona, and
New Mexico (Lomolino and Smith 2003; Service 2007). Over the past 2
centuries, habitat conversion for agriculture, concurrent measures of control,
diseases, competition with cattle ranching operations, and most recently urban and
suburban development have reduced the geographic range of the BTPD to less
than 2% of their original abundance (Miller et al. 1994; Boulder 2006). The
introduced sylvatic plague (Yersinia pestis), kills up to 99% of prairie dogs in
infected colonies (Biggins and Kosoy 2001). In Colorado, about 200 years ago,
prairie dogs inhabited about 2,800,000 hectares. In 2004, the occupied prairie
dogs area was 256,000 hectares. This is almost 11 times less area of their former
geographic range (Proctor et al. 2006). Many remaining colonies are strongly
affected by habitat fragmentation and urbanization (Lomolino and Smith 2001;
Johnson and Collinge 2004). Today, the major reasons for fragmentation and
decline of prairie dog colonies are shooting and poisoning of prairie dogs by
ranchers and agricultural conversion of habitat (Miller et al. 1994).

Habitat requirements
Prairie dogs increasingly compete for available habitat with human
development along the Colorado Front Range. In Colorado, as elsewhere, BTPD
inhabit short and mixed-grass prairies. Short-grass sites are dominated by:
western wheatgrass {Agropyron smithii), blue grama (Bouteloua gracilis), buffalo
grass (Buchloe dactyloides), pasture sagebrush {Artemisia frigida), and woolly
plantain {Plantago patagonica). Mixed-grass sites are dominated by: blue grama
{Bouteloua gracilis)), side-oats grama {Bouteloua curtipendula), blazing star
{Liatris punctata), prairie sage {Artemisia ludoviciana), and aster {Aster falcatus)
(Johnson and Collinge 2004; Delting 2006). In shortgrass steppe of northeastern
Colorado, the cover of blue grama ranged from 9% to 16% at colony-sites and
from 16% to 24% at nearby off-colony sites (Bonham and Lerwick 1976). Similar
trends occurred at other shortgrass steppe sites in northeastern Colorado,
southwestern Kansas, and southeastern Colorado (Winter et al. 2002; Guenther
and Detling 2003). Prairie dogs prefer flat gentle areas or (less than 10 percent
slopes) and well-drained soils that are capable of retaining water for burrow
stability. Steep slopes, tall vegetation, highly sandy and poorly drained soils are
unsuitable habitats for prairie dogs. The species occurs in soil textures ranging
from clays to sandy loams and are attracted to disturbed soils, such as livestock
watering sites and old homesteads (Proctor and Haskins 1998; Proctor et al.

2006). Prairie dogs prefer to colonize areas that have been intensively grazed by
cattle or that have man-made roads and trails. Areas used for cattle are often the
best habitats for prairie dog (Ulev 2007). These areas have shorter vegetation than
ungrazed areas, and therefore require less removal of tall vegetation by prairie
dogs during the initiation or expansion of colonies. Predation at grazed colony-
sites is lower than predation at ungrazed colony-sites for the prairie dogs due to
shorter vegetation and good visibility (Delting 2006; Ulev 2007).
The decline of the BTPD is threatening biodiversity in the grassland
ecosystem due to negative effects on other species that rely on them and/or their
habitat for their own survival (Miller et al. 1990). The black-footed ferret is an
obligate species in that it feeds exclusively on prairie dogs. Mountain plovers
(Charadrius montanus) and burrowing owls (Athene cunicularia) are strongly
facultative species. These species have suffered declines in areas where BTPD
have declined (Kotliar et al. 1999). Mountain plover, ferruginous hawk (Buteo
regalis), and swift fox (Vulpes velox) have been proposed as candidate species
under the Endangered Species Act (Miller et al. 1994). The listing proposals of
those species cited that their decline was due to the decline of the BTPD, an
important source of food for these animals (Miller et al., 1994). The decline of the
BTPD also disrupted prime habitat for these animals because these species rely on
the BTPD burrowing and grazing activities (Miller et al. 1994).

On 31 July 1998, the National Wildlife Federation petitioned the United
States Fish and Wildlife Service to list the BTPD as threatened throughout its
range under the Endangered Species Act (ESA) (Van Pelt 1999). Listing was
deemed warranted by the U.S. Fish and Wildlife Service, but precluded due to the
long list of other higher priority species. This allows time for individual states to
implement their own conservation and/or protection plan (Van Pelt, 1999). The
species was removed as a candidate for the ESA list in August of 2004 (USFWS
2004). BTPD are not given any protection on private lands. Three states: Kansas,
New Mexico, and North Dakota, require mandatory control. Only four states:
Colorado, Montana, New Mexico, and South Dakota, have partial closures on
poisoning (USFWS 2004; PDC 2007). In Colorado, beginning in September
2001, public lands (federal and state) were closed to the shooting of BTPD year-
round. Private landowners were only allowed to control BTPD in damage
situations on private land without a license (Luce 2003). The BTPD is listed by
Colorado Natural Heritage Program (CNHP) as apparently secure globally (G4)
and rare in the state (S3). The species has CNHP partial tracking status and is
ranked as a species of special concern by Colorado Division of Wildlife (CDOW)
and CNHP (CDOW 2003; CNHP 2006).

Recent Research
The attempts to quantify prairie dog habitat suitability in terms of
measurable habitat variables were Clippinger's (1989) HSI model and also HSI
model, produced by the United States Forest Service for the National Grasslands
(unpublished). The Clippinger's HSI model was developed from a review of the
literature and synthesis of existing information of the BTPD (Clippinger 1989).
The model consolidated habitat use information into a framework appropriate for
field application. The resulting index ranges between 0 and 1 and was intended to
express the suitability of an area based on all relevant characteristics of habitat
quality. The Clippinger HSI model defined functions to express habitat suitability
based on percent herbaceous cover, percent slope, average vegetation height, and
soil type. Habitat Evaluation Procedures previously developed by the U.S. Fish
and Wildlife Service were used to design the HSI model (Clippinger 1989).
The analysis, relating patterns of habitat occupancy (occupied/unoccupied)
at the scale of 30 m Landsat TM pixel was performed in a study by Proctor,
Haskins and Beltz (Proctor and Haskins 1998). The HSI model was created to
identify suitable habitat for BTPD within the Charles M. Russell and UL Bend
National Wildlife Refuges, Montana. The HSI model was based on associations
between observed habitat use and measures of habitat quality within the area
modeled. Areas known to be unsuitable such as open water, forested areas, and

wetlands were eliminated from the data, allowing the modeling process to focus
on habitat relationships of interest. The HSI model was able to make relatively
precise and accurate predictions of prairie dog habitats by focusing more closely
on areas of interest. The HSI model was applied to south Phillips County,
Montana, where data on prairie dog presence was also available.
Gribb and associates (2002) developed HSI models for seven endangered,
threatened, or candidate species, including the BTPD in Albany County,
Wyoming. Topography, geology, soils, climate, vegetation and hydrology
variables were used to create HSI models for each species. Subsequent models
were developed in Arc View Spatial Analyst software's ModelBuilder to predict
suitable habitat for each species. A ranking system from 1 to 3 was employed to
classify variables within each attribute used in an HSI model. The final habitat
overlay maps produced for each species illustrated this ranked order of potential
habitat, where number "3" represented areas with the greatest habitat potential.
Areas identified with a "1" and a "2" represented less suitable potential for each
species. Unlike Proctors (1998) study, no effort was made to determine whether
any of the species were actually living in the areas identified as potential habitat
in this study (Proctor and Haskins 1998; Gribb et al. 2001).
Orth and Kennedy (2001) used a GIS and spatial analysis metrics to
examine the landscape around prairie dog colonies assess the suitability of the

habitat for burrowing owls {Athene cunicularia). Burrowing owls have always
been strongly associated with prairie dog colonies. The researchers intended to
determine how land-use patterns and fragmentation around prairie dog colonies
affect nest-site selection by the owls. To test this prediction, a GIS and spatial
analysis metrics were used to examine the landscape within 1000 and 2500m
radius circles surrounding prairie dog towns in the shortgrass prairie in
northeastern Colorado. The test showed that at the 1000m scale owls nested in
towns in which shortgrass patches were a smaller percentage of the total
surrounding landscape. The researchers suggested that owls may be selecting
more fragmented prairie dog habitat because prey availability maybe higher than
in less fragmented landscapes (Orth and Kennedy 2001).
The distribution of the BTPD habitat on the Comanche National
Grasslands in Colorado was mapped using criteria based on slope and soil type
(Ryke 2005). Potential habitat was classified as areas with both suitable slope and
soil type. Unsuitable habitat was classified as all areas with unsuitable slope or
unsuitable soil type, and low potential habitat was classified as all other areas.
The map was reproduced from geospatial information prepared by the U.S.
Department of Agriculture, Forest Service in 2005. The output map represents the
preferred, marginal, unsuitable and unmapped habitats for the BTPD.

The series study of Wyckoff with co-authorship (2004, 2007)
demonstrated how GIS and remote sensing tools can be applied to facilitate
cumulative effects assessments (CEA) on the environment in the metropolitan
region of Denver, Colorado. HSI models including the BTPD were developed
using the Model Builder tools in Environmental Systems Research Institutes
(ESRI) ArcGIS software. The HSI model was based upon three specific
environmental key variables: vegetation type, slope of terrain, and maximum
elevation. The result of HSI modeling was a habitat suitability map. The map
showed the distribution of the BTPD habitat that meets all required parameters for
this species. The output map illustrated locations of sites that could potentially
support prairie dog populations (Wyckoff et al. 2004; Muller et al. 2007).
The purpose of this study was to develop an HSI model for the BTPD
from readily available environmental data using the modeling capability of
ArcGIS9.2 software.
Specific objectives were to:
1. Identify key environmental variables that affect the BTPD and
determine their importance and how they affect the assignment of
habitat priority.

2. Solicit and obtain expert opinion on weighting of variables in the
BTPD HSI models to further refine output.
3. Select a study area within Colorado for the BTPD where it is known to
occur to develop a model.
4. Create the HSI model and run this model to determine if the model
appears to depict actual sites.
5. Verify the ability to accurately predict HSI model output between
suitable habitats areas and breeding site localities of the species.
6. Provide recommendation for resource managers who can use the HSI
model for BTPD and produce map to identify and prioritize habitats
for management and future monitoring in Colorado.

Study Area
The Comanche National Grassland (CNG) is approximately 178.062 ha
located in southeastern Colorado. GNG is divided into two units, the Timpass unit
(northern portion) and the Carrizo unit (southern portion) (Hazlett 2004). The
Carrizo unit was selected as study area for the BTPD model development because
this area is having numerous BTPD colonies (Fig.2.1). The Carrizo is primarily in
Baca County, in the general vicinity of town of Springfield, and extends into
extreme eastern Las Animas County (Mackessy and Stephen 1998).
Administration of these National Forest System (NFS) lands is by the United
States Department of Agriculture (USDA). The elevation range in the CNG from
1,219 m in southeastern Baca County is to 1,829 m on Fallas, Tecolote, and
Carrizo Mesas areas. In the CNG, 70 percent to 80 percent of the annual
precipitation occurs as rainfall between April and September (Mackessy and
Stephen 1998; Hazlett 2004).

Figure 2.1. Map of the Carrizo study area
in the Comanche National Grassland, Colorado
Southeastern Colorado is a sparsely inhabited region roughly 233.4 km by
217.3 km (nearly 51,800 km ). Much of this area is shortgrass and mixed grass
prairie. Common species in the area include: sego lily (Calochortus gunnisonii),
red paintbrush (Castilleja Integra), pinyon pine (Pinus edulis), and Ponderosa
pine (Pinus ponderosa). In the Southern sections of the plant species more
characteristic of the Chihuahuan sub-province are found, including cooley
bundleflower (Desmanthus cooleyi) and pink mimosa (Mimosa borealis) (Hazlett

2004; Ulev 2007). The open step of CNG land is represented by buffalo grass
(Buchloe dactyloides) and galleta grass (ilaria jamesii) are common on fine-
textured soils, blue grama (Bouteloua gracilis) is more common on loam soils,
and western wheatgrass (lymus smithii) is more abundant in mesic swales. In
some steppe areas snakeweed (Gutierrrezia sarothrae) shrubs are common. In
other areas cholla or cane cactus (Opuntia imbricata) is conspicuous on the steppe
landscape (Gillihan and Hutchingsm 1999; Hazlett 2004). Mixed-grass and
tallgrass communities in Baca County are locally heavily overgrazed with
extensive fecal contamination (Mackessy and Stephen 1998). In addition to
shortgrass prairie, extensive canyonlands habitat is found in CNG. Large areas of
agricultural land intermingling with public lands on the Carrizo unit, contributes
to conflicts with wildlife species including prairie dogs.
Spatial Database Development
Geographical information system coverage data were obtained from a
variety of sources including the U.S. Bureau of Land Management (BLM),
Colorado State agencies, and local government entities. All data relevant to the
research were collected and archived into a coherent database for easy retrieval.
To assist in keeping track of the large number of raw spatial data sets, a
Cumulative Effects Data Inventory Dictionary (CEDID) was developed. The

CEDID a Microsoft Access database and provided an easily usable and
centralized method of managing the metadata for different coverages. A 30 m
resolution Digital Elevation Model (DEM) was used with the ArcGIS9.2 Spatial
Analyst extension to derive topographical variables including slope and aspect.
All GIS coverages used for the HSI modeling are described in Appendix A. Data
include vegetation, slope, hydrology, and transportation networks relevant to the
study area. In the project the available soils data was deemed to not be compatible
in its original form and that conversion to something useful would have been too
costly for your project. The vegetation coverage for the BTPD HSI study sites
was obtained through a project conducted by the BLM, the Colorado Division of
Wildlife (CDOW), and the Colorado Vegetation Classification Project (CVCP).
CVCP utilized Landsat Thematic Mapper image data, ERDAS MAGINE9.1/8.3
and Arclnfo software. Each vegetation polygon represents unique
combinations of cover for several dominants species of grasses or forbs, shrubs
and bare ground. The vegetation layer was analyzed for its ability to spatially
characterize vegetation with respect to species composition associated with BTPD
habitat requirements. Land cover type categories are based on a modified
Anderson Land Use and Land cover Classification System (1977). Road network
data and hydrology data were obtained from the U.S. Census Bureau 2000

TIGER/Line Data (ESRI 2000). The range of the BTPD was clipped to the
boundary of the Carrizo unit.
Habitat Suitability Mapping
Habitat for key indicator species and ecosystems was determined using the
HSI approach, which maps the suitability of lands for habitat and environmental
value. HSI mapping builds on research from a number of other studies including
the Habitat Evaluations Procedure (HEP) (USFWS 1980). The procedure was
developed and established by the USFWS to provide a standardized process for
modeling wildlife habitats. Using the modeling approach, habitats are assigned
relative values depending on how well they provide the requisites. HSI models
incorporate basic life requisites such as a food, land cover, slope and soil type.
Life requisite values are then combined to obtain on HSI for a specified species
(Donovan et al. 1987). The flow chart of procedure approach for solving spatial
problems in HSI modeling for BTPD is illustrated in Figure 2.2. In this study, the
HSI model was developed for the BTPD based on the HEP procedures describing

Figure 2.2. Approach of solving spatial problems for the HSI model for BTPD

Weighted Overlay Technique
Weighted overlay is a technique for applying a common scale of values to
diverse and dissimilar input data to create an integrated analysis (ESRI 2007). The
Weighted Overlay Tool in ArcGIS software allows all inputted values to be
taken into consideration. It reclassifies values in the input raster into a common
evaluation scale of suitability or preference, risk, or some similarly unifying scale.
The input rasters are weighted by importance and added together to produce an
output raster. The model was developed using the Model Builder tools in ESRI
ArcGIS software. ESRIs Software was used for the weighted overlay approach
and repeated for all categories in each map. The steps are summarized below
(ESRI 2007):
A numeric evaluation scale is chosen. The end-points on the chosen
scale represent extremely unsuitable and extremely suitable.
Between these extremes there lies a continuum of conditions.
The cell values for each input theme in the analysis are assigned
values from the evaluation scale and reclassified to these values.
This makes it possible to perform arithmetic operations on grids that
originally held dissimilar types of values.
Each input theme is weighted (i.e. assigned a percentage influence
based on its importance to the model).


f \
Input to
Input to

Figure 2.3. Steps involved in HSI model for BTPD in Carrizo study area

The total influence for all themes equals 100 percent.
The cell values of each input theme are multiplied by the themes
The resulting cell values are added to produce the output grid theme.
The HSI model relies on habitat preferences of BTPD and environmental
data, and the use of weighted overlay and other functions that characterize the
suitable species habitat. Figure 2.3 illustrates the HSI model of BTPD. As you can
see from Figure 2.3, a number of factors including vegetation, hydrology, slope,
and road network were carefully considered. Each input was used to evaluate
BTPD habitat suitability. For instance, the BTPD prefers flat and open
topography, so these areas were rated as a most suitablethan those with moderate
to steep slope. Prairies dogs can create patches of a unique habitat of bare ground
and short, sparse vegetation over time (Lauenroth and Milchunas 1992; Winter et
al. 2002; Winter et al. 2003). Bare ground and short, sparse vegetation are
represent the most suitable habitat of BTPD (Hoogland 1995; Proctor and Haskins
1998; Ulev 2007) and were produced a high habitat rating for the BTPD HSI
model. The process of habitat classification was repeated for all six categories in
the HSI map of the BTPD. Once all overlays are complete, those areas with the

highest overall habitat suitability will be identified and will be marked in red. In
contrast, the deepest green was representing very unsuitable areas for the BTPD
where one would not expect to find any BTPD habitats. The overlay techniques
that were used in this study were completed in GIS.
Model Construction and Mapping
The HSI model for the BTPD was developed using the geospatial
modeling capabilities of ArcGIS9.2 software. A raster GIS was used to store,
manage and analyze data needed in the suitability analysis and display the result
of these analyses. Each habitat factor was converted to a raster format layer in
order to technically enable habitat suitability analysis in a GIS environment.
These map layers describe the spatial variation of each habitat factor in a certain
area. The HSI model was constructed based on habitat requirements for the BTPD
habitat as described in the peer-reviewed literature and including personal
contacts with species experts in Colorado. The modeling procedure was based on
habitat variables, which cover the most essential characteristics of species
preferred habitats. The boundary of the study area was clipped according to the
occupied range of the BTPD habitat in Colorado and then clipped within the study
area in the Carrizo unit in the CNG. Since, the available data from CVCP
(vegetation map with original 30 m pixel resolution) were used as a primary data

to classify the BTPD habitat; distribution maps for this species were generated at
a cell ground resolution of 30x30 m. Habitat suitability scores were selected after
a thorough literature review of related studies, and expert interviews. The HSI
scores were based primarily on habitat preferences for the BTPD and are
influenced by vegetative cover type, slope (Reading et al. 1989).
Using the South West (SW) Re GAP land cover (vegetation) descriptions
(Lowry 2005b; Lowry 2005a; Lowry et al. 2007; NatureServe 2007), Nature
Serve 2004), vegetation was categorized as suitable or unsuitable for habitat
areas. The SWReGAP vegetation layer was clipped to the area under study. A
table was created for the various vegetation types and all land types unsuitable to
the BTPD were given a pixel value of 0 as the species is not known to occur with
that particular habitat type. All other types of vegetation were then rated on a
scale from 1 to 5, where it was assumed that class 5 habitats were the most
suitable and class 1 were the least suitable. The table was used to reclassify the
SWReGap vegetation land types raster so that all land areas were given a unique
pixel value from 0 to 5. The BTPD is commonly associated with short-and mixed-
grass prairie, sagebrush steppe, and desert grassland, so the highest pixel values
were given to these vegetation types. Short grass prairie dominated by buffalo
grass (Buchloe dactyloides), blue grama (Bouteloua gracilis), and western
wheatgrass (Pascopyron smithii), and mixed-grass prairies that have been grazed

by native and nonnative herbivores are the preferred habitat for BTPD (Bonham
and Lerwick 1976; Clippinger 1989; Fahnestock et al. 2003; Ulev 2007).
According to the literature, BTPD live between 700 m and 1,700 meters
(Hoogland 1995) and prefer slopes less than 10 % (Tileston and Lechleitner 1966;
Clippinger 1989; Hoogland 1995; County 2002). A slope from 0 percent o 5
percent is high potential and a slope from 5.1 percent to 10 percent is low potential
for BTPD habitat. Slope as an indicator of terrain steepness was determined using
a USGS DEM of 30 m spatial resolution. Using ArcGISs spatial analysis tools a
slope from DEM data. The elevation slope file was clipped to the BTPD study
area and then reclassified on a six class-ranking scheme (Table 2.1).
Table 2.1 Slope variables and habitat classes within the study area of BTPD
Slope range, % Habitat Class
0-2% Very High
2-4% High
4-6% Medium
6-8% Low
8-10% Very Low
>10% Unsuitable
The highest pixel values were given to slope range from 0-2 percent as
the most suitable areas and the lowest pixel values were assigned to slope range
from 9-10 percent as less suitable habitats and pixel values of 0 were given to
slopes greater than 10 percent. By removing unlikely categories from

consideration, remaining possibilities would be more accurately defined as
suitable or unsuitable. GIS rasters, vegetation, and slope, were reclassified on
common measurement scale of 1 to 5, being weighted using the Weighted
Overlay Approach. This is a multiplicative approach in which all layers were
combined to create a single output map.
The weighted overlay table allows the calculation of a multiple criteria
analysis between several rasters. The influence of the raster was compared to the
other criteria as a percentage of 100. Each raster was assigned a percentage
influence according to the habitats importance for the BTPD. Vegetation types
were given the greatest weight 60 percent in the model value equal of the
variation and slope 40%. Cell values are multiplied by their percentage influence,
and the results are added together to create the single output raster of the entire
area. Areas on the output map with the highest overall habitat suitability were
depicted in bright red. Deep green colors represented unsuitable areas for the
Low potential BTPD habitat on the grasslands is represented by areas
where soils, slope, and vegetation are generally limiting to prairie dog occupancy
(Clippinger 1989). BTPD colonies inhabit alluvial soils with medium to fine
textures, and occasionally gravel. Soil that is not prone to collapsing or flooding is
preferred by the BTPD (Clippinger 1989; Augustine et al. 2007). The potential

surface soil textures for BTPD colonies on the CNG with slopes less than 5
percent are: loamy uplands, loamy plains, limey uplands, alkaline plains, loamy
bottomlands, basalt loam and clayey Low potential types of soil with slopes from
5.1 to 10 percent include: sandy plains, gravelly breaks, saline overflow, playa,
salt flat, gravel/eroded, limestone and shaley plains (Ryke 2005). Unsuitable for
BTPD habitats soil types represent by sandy bottomland, choppy sand, deep sand,
sandstone breaks and basalt breaks with slope greater 10 % are (Ryke 2005).
The BTPD are tolerant of vehicles at close distances, but can show
changes in behavior if people get out of their vehicles (Hoogland 1995). Highway
layers were clipped to the study area, and buffered by 15 m on each side and then
were converted to a raster grid. The road raster grid was reclassified from 0 to 1,
where all highways were given a value of 0 and listed as unsuitable habitat areas.
The BTPD species does not associate with temporary or permanent water
bodies. They obtain water from plants (Clippinger 1989). Open water reservoirs
were selected as unsuitable key variables for BTPD habitat in the HSI model. The
hydrology layer was clipped to the study area and then converted into a raster file.
The hydrology raster grid was reclassified from 0 to 1, where pixel value of 0 was
listed as unsuitable area and a pixel value of 1 as suitable habitat. The reclassified
hydrology and road rasters were multiplied on a cell-by-cell basis to exclude the
unsuitable habitat. To create the HSI model, the single landscape suitable raster

was multiplied by the unsuitable raster on a cell-by-cell basis. The HSI model
with all required parameters for the BTPD was run. The output layer was
combined to create an overall suitability map, which represents locations in the
study area that contain suitable habitat for BTPD.
Methods of Validation for HSI Model
Evaluating the effectiveness of an HSI model typically entails using
independent data to test the ability of the model to accurately predict occurrence
and or abundance of target species. Although the data of site locations surveys
was not explicitly used to generate habitat suitability models, the peer-reviewed
literature and including personal contacts with Dr. John Wyckoff, who is an
expert of prairie dogs, were used as sources of information in selecting model
variables and classification criteria. Accuracy of the model was examined by
overlaying GIS species occurrence layers onto predicted habitat suitability map.
Databases of known BTPD colonies in the State of Colorado were selected
to independently test the habitat quality model. These data were compiled from
existing BTPD databases varying in date and scale by Colorado Bird Observation
(CBO) biologists field. The database allows colonies to be queried on their status,
date last visited, associated species and the source of the information. The
database represents a compilation of accessible mapped information. The data

were published by University of Colorado at Denver and Health Sciences Center
from Arclnfo coverage created by EDAW Inc., Fort Collins, Colorado office on
August 09, 2000. Fred Nuszdorfer from the University of Colorado, Denver,
Department of Civil Engineering compiled the database (Johnson et al. 2006).
The latest available updated dataset, containing 44 BTPD colony sites for the year
2000, (Johnson et al. 2006)was selected to independently test the habitat quality
model and identify the most important variables in relating BTPD presence to
habitat suitability within the Carrizo study area. The capabilities of Spatial
Analyst tool of ArcGIS9.2 software were utilized to calculate the HSI category
for each separated colony site. The HSI layer was clipped by the single layer of
BTPD site colonies to obtain the layer of colony sites with six habitat zones. The
procedure was repeated 44 times for each colony sites. Microsoft Office Excel
2003 software was used to calculate the habitat area in ha for each separated
layers of colony sites with six habitat zones.
The Summarizing Statistic approach was used to examine whether the
depicted frequencies of occurrence in each habitat category was different than
would be expected if the number of colony location sites simply reflected the
availability of each predicted habitat category. GIS data of the BTPD colonies for
the 2000 year were used to verify the ability to accurate predict the HSI model
output between six habitat categories for the breeding site localities of the species.

Microsoft Office Excel 2003 software was used to calculate in percent how
many colony site locations fall within with each of the six categories. One-way
ANOVA tests with equal variances and Tukeys multiple comparison test were
used to determine if the differences between the habitat areas, occupied by BTPD,
are significant (Bahn 1998; North et al. 1999; Burger and Bahn 2004;
Gillenwatera et al. 2006). A one sample two sided t-test was used to analyze the
difference between the habitat areas with the hypothesis that the mean of the
differences (0.488) equals unsuitable habitat due to unequal variances between
Medium and Unsuitable habitat areas All statistical analyses were carried out in
Minitab15.1.1.0 (Bahn 1998; Gillenwatera et al. 2006).
Sensitive Analysis Approach
Sensitivity analysis is used to determine how sensitive a model is to
changes in the value of parameters of a model and to changes in the structure of a
model (Crosetto and Tarantola 2001). Parameter sensitivity is performed a series
of tests. For example, modeler sets different parameter values to examine how
sensitive the choices are to changes in criteria weights. Sensitivity analysis is a
useful tool in model building (Crosetto and Tarantola 2001). This method allows
to exam how models behavior responds to changes in the parameter values
(Crosetto and Tarantola 2001; Store and Kangas 2001). Additionally, sensitivity

analysis allows building confidence in the model by studying the uncertainties.
The uncertainties usually are associated with parameters in HSI models. Results
of sensitivity analyses can show which variables have a large impact on HSI
models output (Crosetto and Tarantola 2001; Store and Kangas 2001; Store and
Jokimaki 2003). Also, sensitivity analysis can indicate which parameter values are
reasonable to use in HSI models (Crosetto and Tarantola 2001; Store and Kangas
2001; Store and Jokimaki 2003).
The purposes of sensitivity analysis in this study were: to find out the
influence of different criteria weights on the spatial pattern of the BTPD HSI
model and determine how sensitive the model to these variables. To exam the
effect of the changing weighted variables in the BTPD HSI model, to the
vegetation and slope were given different weighting schemes. Weighting schemes
were applied for vegetation and slope characteristics as follow: 0.5 and 0.5, 0.7
and 0.3, 0.8 and 0.2, 0.9 and 0.1, and 1.0 and 0.0. Suitability maps for every
weighting scheme were created in GIS. In suitability maps the pixels of the index
maps were classified into six categories. Sensitivity analysis this study was done
by comparing BTPD suitability maps.

Validation of HSI Model
Six habitats of BTPD depicted on the HSI map (Fig 3.1) are scattered
throughout the Carrizo study area. Within the study area BTPD occur primarily in
the shortgrass prairie ecological area. The majority of Very High and High
mapped habitats, depicted in red and orange color, occur in the western and
northern regions of the park with isolated patches throughout the southern and
western boundaries (Fig. 3.1). Within the Carrizo study area, approximately 92%
of the land is potential habitat for BTPD and was occupied by 92 % of the BTPD
BTPD suitable habitat areas represent by shortgrass prairie, grassland and
forbland Landcover (EDAW 2000; EDOW 2000; Ryke 2005). Large amounts of
potential habitat for the BTPD in sandsage areas on private is converted to
agricultural uses (Ryke 2005). As we can see from the HSI map for the BTPD
(Fig. 3.1), the largest block of suitable area located in the northwest side of the
Carrizo unit has not been occupied by the BTPD colony as would be expected.
This area might be privately owned land, and poisoning and shooting of BTPD by
rancher as well as agricultural conversion of habitat are likely explanations for the

majority of the decline of species. Furthermore, the introduced sylvatic plague
(Yersinia pestis), could have killed up to 99% of prairie dogs in infected colonies
in this region (Miller et al. 1990; Miller and Knopf 1993; Miller et al. 1994;
Hoogland 1995; EDOW 2000; Miller et al. 2000; Biggins and Kosoy 2001; Cully
et al. 2001).
Medium habitat occurs in the southern region of the park and is composed
of semi-desert shrub steppe, agricultural lands, and lands that have recently
burned. The Low and Very Low habitats in the study site represent areas where
vegetation consists of sandhill shrubland, wood shrubs, mixed low sagebrush
shrubland, and salt desert scrub. Areas with slopes less than 6-8% are generally
within the limits of BTPD occupancy (Fig. 3.1). The presence of woody shrubs
such as Artemesia filifolia impedes visibility and sandy soils make it difficult for
prairie dogs to dig, however, patches of loamy soils are capable of supporting
small BTPD colonies. Additionally, current vegetation height, due to lack of fire
or low grazing pressure, may limit BTPD occupancy in these areas. Unsuitable
habitat represents areas where vegetation, soil, slope, roads and lakes are
generally preventing occupancy by prairie dogs. Potential BTPD habitat
represents areas where habitat disturbance is limited through management
processes including fire and grazing, and population regulation factors such as
disease, predation, land use on adjacent private lands, fence and road densities,

invasive plant species and dispersal could have the greatest effect on BTPD
distribution, fragmentation and decline (Clippinger 1989; Hoogland 1995; Delting
1998; EDOW 2000; Biggins and Kosoy 2001; Boulder 2006; Delting 2006).
As can be seen from the HSI map (Fig.3.1) for BTPD, the distribution of
potential suitable areas throughout the CNG are highly fragmented: few large
blocks of BTPD data for the year 2000 on the Carrizo study site showed that most
colonies occur in Very High and High potential habitats, while few colonies occur
in the Low potential or unsuitable habitats (Table 3.1). One of the reasons for the
fragmentation of suitable areas in the Carrizo Unit is the high percent of boundary
length (66%) and high percent of number of fragments (90%) to the whole area
in CNG (Ryke 2005). Fragmented ownership of the land is problems in studying
and managing the population of wildlife species, including BTPD population
(Ryke 2005). On private lands, prairie dog presence conflicts with agricultural or
cattle production objectives. Many landowners strive to eradicate prairie dogs on
their lands (Clippinger 1989; Delting 1998; Ryke 2005; Boulder 2006; Delting

Habitat Suitability Index: Black-Tailed Prairie Dog
The Comanche National Grassland, Carizo unit. Colorado
Figure 3.1. Output map from HSI model for the BTPD

As we can see from Table 3.1 Very High and High suitable habitats
contain 44 % and 37 % of BTPD habitat suitable areas and were inhabited by 92
% of the BTPD colonies. According to the data, we can conclude that Very High
and High suitable habitat represented the most preferred habitats of BTPD
colonies. Medium suitable habitat also represents potential BTPD habitat with a
14 % occupancy rate and 8 % of existing colonies. Low and Very Low habitats
represent potential habitat areas with less than 1% suitable lands within the habitat
areas and only 0.05 % of existing PTBD. Within the study site, only 13,601 ha
fall within Unsuitable habitat. This equals about 4 % of the total area. In
comparison, 32,4612 ha, or 82 %, of the study site falls within the most preferred
habitat. Low and Very low suitable habitats represent only 2,385 ha or less than
1% of the study site. Analysis of the summarizing statistics shows a strong
correlation between occupied BTPD colony sites in the year 2000 and the HSI
BTPD model output, which indicates that the HSI model works and also that the
majority of land of the Carrizo Unit is potential suitable habitat for BTPD.

Table 3.1 Habitat suitable areas based on the output of HSI model
in the Comanche National Grassland, Carrizo unit
HSI area Habitat Suitable Areas based on the HSI model Area, occupied by 44 BTPD colonies in the year 2000
Hectares (he) Percents, % Hectares (he) Percents, %
Unsuitable 13601.79 4.02 1.35 0.08
Very Low 293.49 0.09 0.00 0.00
Low 2092.32 0.62 0.81 0.05
Medium 45721.44 13.52 126.45 7.57
High 126357.39 37.36 602.91 36.09
Very High 150147.45 44.39 938.97 56.21
TOTAL 338213.88 100 3340.98 100
Results of one-way ANOVA with equal variances and Tukeys multiple
comparisons tests were used to determine if the differences between potential
suitable areas occupied by BTPD in Very High, High and Medium habitat areas
were significant (Bahn 1998; Gillenwatera et al. 2006). A one sample two sided t-
test was used to analyzed the difference between the habitat areas (Medium and
Unsuitable) with the hypothesis that the mean of the differences (0.488) equals
unsuitable habitat due to unequal variances between these habitat areas. Both tests
have been performed using Minitab15.1.1.0. The output of the statistical tests is
provided below.

Table 3.2 One-way ANOVA Test (Percent vs by Zone)
for the HSI model for BTPD
Source DF Sum of the Squares Mean Square F Ratio P
Low habitat area 2 440.75 220.38 33.11 0.000
Error 129 858.55 6.66
Total 131 1299.30
S = 2.580 R-Sq = 33.92% R-Sq(adj) = 32.90%
Table 3.3 Individual 95% CIs for Mean Based on Pooled StDev
HSI Area N Mean StDev + + +
Medium 44 2.499 1.960 ( *___)
High 44 5.029 2.918 ( * )
Very 44 6.962 2.758 ( *- )
3.2 4.8 6.4 8.0
Table 3.4 One-Sample T: Zone 3. Test of Mean=0.448 vs not= 0.448
HIS Area N Mean StDev SEMean 95% Cl T Ratio P
Medium 44 10.00 16.07 2.42 (5.11, 14.89) 3.94 0.000
The Tables above display convincing evidence that the percent of habitat
area occupied by BTPD in Unsuitable habitat area is less than that of Medium
habitat area as confirmed by a two-sided one-sample t-test (PO.OOl). Moreover,
there is convincing evidence that percents of suitable habitat areas occupied by

BTPD are different for Very High, High and Medium habitat areas (Table 3.3).
This difference is statistically significant since one-way ANOVA gives P<0.001.
The percent of suitable habitat areas in Very High is 1.38 times greater than in
High habitat area, the percent in High habitat is 2.01 times greater than in
Medium habitat, and the percent of area in Medium habitat is 5.58 times greater
than in Medium Unsuitable habitat area. As we can see from the Table 3.1,
approximately 92 % of all BTPD colony sites in the year 2000 occurred in High
and Very High and 14 % in Medium habitat suitability areas. This equals about
95% of the total study area. Fewer BTPD individuals occurred in Low and Very
Low habitat and Unsuitable habitat areas than expected. Hence, according to the
results of statistical analyses, we can conclude that we do have a good model for
suitable prediction of BTPD habitat in Carrizo study site of Comanche
Sensitivity Analysis
For the purpose of sensitivity analysis, HSI maps (Fig. 3.2) for every
weighting scheme were created using the geospatial modeling capabilities of
ArcGIS9.2. Sensitivity analysis was accomplished by applying different
weighing schemes for two main decision criteria: vegetation and slope. The
weight of vegetation (0.6) and the weight of the slope (0.4) demonstrate the

importance of two main factors (Wyckoff et al. 2004). In this study weighting
schemes were applied for vegetation and slope characteristics, respectively: 0.5
and 0.5, 0.7 and 0.3, 0.8 and 0.2, 0.9 and 0.1, and 1.0 and 0.0. Table 3.5 represents
the results of how changes in weighting affect the choice of habitats to be
conserved. Sensitivity analysis, where weights 0.5, 0.7 and 1.0 were applied for
vegetation characteristics for HSI maps of the BTPD are illustrated in Figure 3.2.
Table 3.5 Sensitivity analyses, where weights 0.5, 0.6, 07, 0.8, 0.9 and 1.0
were applied for vegetation characteristics
HSI Zone W =0.5 Area, % W = 0.6 Area, % W = 0.7 Area, % W = 0.8 Area, % W = 0.9 Area, % W= 1.0 Area, %
Unsuitable 4.02 4.02 4.02 4.02 4.02 4.02
Very Low 0.09 0.09 0.09 0.14 0.24 0.24
Low 0.62 0.62 0.66 2.38 7.88 7.88
Medium 13.52 13.52 13.18 45.75 40.15 40.15
High 37.36 37.36 37.66 1.28 0.40 0.40
Very High 44.39 44.39 44.39 46.43 47.32 47.32
Figure 3.2 shows that the most suitable habitat areas (Very High and
High) are highly fragmented across the study. As the importance of vegetation
characteristics approached increased 1.0 (Table 3.5), then the location of these
suitable areas shifted to the western side of the park. Within the weighed
vegetation range of 0.56 0.7, variations in all six-habitat areas did not fluctuate.

Sensitivity analyses show that, in the case of the BTPD, the vegetation factor had
a major influence on the results of the HSI model (Figure 3.3). According to the
statistical analysis strong correlation was found between suitable habitat areas
GIS data of BTPD colonies in the year 2000. Validation of the HSI model
provides evidence that the model works (PO.OOl). Accordingly, key variables,
including land cover class, elevation limits, slope, and hydrology as they apply to
the respective species, were selected and reclassified correctly. Sensitivity
analysis verified the ability to accurately predict the HSI model (Figure 3.3)
output between the suitable habitat areas and the breeding site localities of the
species. However, natural habitat can be suitable even without the species present,
so a few blocks of unoccupied by BTPD colonies in the year 2000 are displayed
on the HSI map (Figure 3.1). The HSI model (Figure 3.3) servers an important
function for conservation biologists and land managers concerned with preserving
the BTPD diversity in the Carrizo study site at Comanche National Grassland,

W = 0.5
W= 0.7
W = 1.0
Figure 3.3 Sensitivity analyses, where weighs 0.5, 0.7 and 1.0
were applied for vegetation characteristics

Figure 3.3 The ModelBuilder flow diagram for the for the HSI model of the BTPD
Key variables: LCover-Vegetation layer, Reclasswregap table of vegetation type, Elev(2) and
Elev Elevation layers, LakesCl hydrology layer and HW road network layer

The main goal of the study was to develop the HSI model for the BTPD
from readily available environmental data using the modeling capability of
ArcGIS 9.2 software and identify the potential suitable distribution for BTPD in
Carrizo study area of CNG. The BTPD HSI model was based on datasets derived
from remotely sensed imagery and publicly available data acquired from various
agencies and expert opinions (Wyckoff et al. 2004). The HSI model for the BTPD
has not predicted habitat suitability over time, but as element conditions and
distributions are updated, the model is accurately predicting suitable habitat under
current conditions. The HSI model for BTPD incorporates spatially explicit
variables and advances the understanding of wildlife-habitat relationships.
The advantage of HSI modeling is the transition of existing knowledge of
species habitat requirements into quantitative measures of habitat quality In
addition, HSI modeling allows for the connecting of possibilities to consider
habitat factors on different scales (USFWS 1981b). The methods utilized in the
present study demonstrate the possibility to produce suitability habitat indices for
large areas within a reasonable period of time. Determining habitat factors based

on the species preference and their weights was one of the important stages in the
HSI modeling process. The result of an HSI model can have error even if slight
changes in weight coefficients occur. In addition to statistical evaluation, it is
possible to valuate HSI models by using only expert knowledge instead of
statistical evaluation models based on empirical data (Store and Kangas 2001;
Store and Jokimaki 2003).
This study has demonstrated a number of techniques including overlays,
reclassifications, and weighting scales for developing the HSI model for the
BTPD. The GIS-based development HSI model for BTPD in Carrizo unit,
including evaluation approaches, has demonstrated achievement of the original
objectives of this study. The research of habitat requirements of BTPD allowed
for transformation of the key habitat variables into spatial GIS database relevant
to species using ArcGIS9.2 Software capabilities. The methods presented in
this study were based on analyzing and overlaying the data in raster format. The
key variables related to the BTPD habitat covers the most essential habitat
characteristics of BTPD preferred habitats as described in the peer-reviewed
literature and including personal contacts with species experts, J. Wyckoff.
Development of the suitable index based on the requirement factors of the
BTPD habitat was used to identify the potential suitable lands within the study
site. The GIS data set of the 44 BTPD colonies for the year 2000 was utilized to

verify the ability to accurately predict the HSI model output between the
suitable habitats areas and the breeding site localities of the species.
Results of validation tests of the HSI model showed reasonable
correspondence with mapped BTPD colonies and provide evidence that the model
works (P<0.001). Sensitivity analysis in this study was applied to investigate the
effects of changes in the weights of BTPD habitat factors. In addition to the
uncertainties associated with the weight coefficients, the quality of the final
results was influenced by the accuracy of the input data and calculation error. In
the case study of BTPD, sensitivity analyses show that vegetation factor had a
major influence on the results of the HSI model and within the weighed
vegetation range of 0.56 0.7 output variations of six predicted habitat categories
did not fluctuate. Due to the literature review of recent publications on BTPD
habitats (Winter et al., 2003; Wyckof et al., 2007; Ryke 2005) it was possible to
incorporate the most up-to-date information on BTPD preferred habitats into a
spatial model of BTPD habitat suitability. The model accounts for all known
BTPD habitat preferences and calculates a weighted average suitability value for
each of the six habitat areas in the study area. Since, the vegetation maps with 30
m pixel resolution, created by CVCP, were used as a primary data to classify the
BTPD habitat, the model measures variation of habitat criteria across space and
quantify suitability for BTPD developed at the finest possible resolution of 30m.

The use of this model allows the wildlife manager to develop the most
effective, spatially targeted programs to protect suitable BTPD habitat. In future
research, this model could be expanded upon to include an analysis of
fragmentation and clumps of suitable habitat areas across years. Additional,
analyses should also include a comparison of preferable habitat areas to locations
that are at risk of being destroyed or developed. An important implication of the
HSI modeling is that a high quality habitat may play a consequent role in assisting
isolated breeding sites. Protecting sites where high quality habitat is abundant
should positively influence the persistence of local populations (Service 2007).
This is very important for the BTPD colonies have had patchily distributed sites
of high quality habitat that were often isolated from one another by relatively
large distances.
The HSI model, using remote sensing and GIS data, provides important
information for conservation biologists and land managers concerned with
preserving the BTPD (Service 2007). The HSI model that was produced in this
project is a source of information in the environmental planning process for
predicting the impacts of resource management practices on wildlife habitat.
HSI model can be used as a tool to identify sites for future surveys as areas of
high species abundance. The optimistic result of the BTPD HSI model will be
very useful for conservation partners who can focus on potential high-quality

habitat areas to restore habitats or protect those important to the BTPD within
the Carrizo study site at CNG. The model can also be used to identify new
potential management areas. Resource managers can use the HSI model and
map to identify and prioritize habitats for conservation actions and future
monitoring of BTPD on the CNG, Colorado. Resource managers may use future
extensions of this habitat modeling as a basis for evaluating threats to breeding
habitats. The developed HSI model can be applied only to the BTPD, and not to
any of the four species of Cynomus because other three species has different
habitat preferences. For example, the white-tailed and Gunnison prairie dogs
have loosely social towns of lower density and lesser impact on vegetation,
which is ecologically has differences from the dense of BTPD that prefer short
vegetation (Hoogland 1995). The BTPD model could be used in making
decision on prairie dogs and the species that depend on prairie dogs as prey.
Possible uses of this model include the evaluation of habitat suitability of
current colony sites, the evaluation of possibilities for colony expansion, and the
suitability sites for transportation or rehabilitation of prairie dogs. The detailed
analysis of alternatives that can be developed for possible implementation of the
BTPD management strategy can be found in reports of the ED AW (2000), City
of Boulder (2006), Ryke (2007), and National Park Service (2007).

Appendix A
Datasets of GIS coverages used for creation the HSI model of BTPD in the
Carrizo study area, the Comanche
Layer name Description of dataset Data Management
Grid Coordinate Universal Transverse Clipped to the study
System: Mercator area Created a
Projection Transverse Mercator separate layers of the Carrizo unit in the
Projected NAD 1983 UTM Zone Comanche
coordinate system 13N
Geographic GCS North American 1
coordinate system 983
Data ShapefilePolygon
FormatFeature Type: Location of the Data /Access Date Colorado University at Denver Server=CE-NC2406-
SDE.ESRIstr Grid Coordinate
eetsUTM System:
coordinate system
coordinate system
Universal Transverse
Transverse Mercator
Clipped to the
Carrizo Boundary
Creates buffer
polygons to a 15 m
around the road
network Converted to
the Raster, and
reclassify from 0 to 1,
where for all roads
were given value of 0
listed as unsuitable

Location/Access Date Colorado University at Denver Server=CE-N C2406-; Service=esri_sde; Database=cuacel; User=sde; Version=SDE.DEFAULT (download from U.S. Census Bureau 2000 TIGER/Line Data (ESRI 2000). 02/11/2008
Description: Road network data represents detailed streets, interstate highways, and major roads within the Colorado State
SDE.StateEng Lakes Grid Coordinate System: Projection Projected coordinate system Geographic coordinate system Universal Transverse Mercator Transverse Mercator NAD 1983 UTM Zone 13N GCS North American 1 983 Clipped to Carrizo Boundary Converted to the Raster, and reclassify from 0 to 1, where for all lakes, reservoirs and pond were given value of 0 listed as unsuitable
Data FormatFeature Type: ShapefilePolygon areas
Location/Access Date Colorado University at Denver, Server=; Service=esri_sde; Database=cuacel; User=sde; Version=SDE.DEFAULT doenload from U.S. Census Bureau 2000 TIGER/Line Data (ESRI 2000). 02/15/2008
Description: Hydrology data represents detailed water reservoirs, lakes and ponds within the Colorado State

z040126demfi Grid Coordinate Universal Transverse Clipped to Carrizo
1 System: Mercator Boundary Derived
Projection Transverse Mercator slope and curvature indices with ArcGIS
Projected NAD 1983 UTM Zone tools
coordinate system 13N
Geographic GCS North American 1
coordinate system 983
Data FormatSource Type: Raster_ Continuous
Location /Access Colorado University at
Date Denver, WCEPHDNC2406- 2\D$\GIS-
Description: Data\dem30m\z040126de mfil 02/15/2008 DEM30m of average elevations
SDE.landcove Grid Coordinate Universal Transverse Clipped to Carrizo
r System: Mercator Boundary Links to
Projection Transverse Mercator the table mpregap.mdb
Projected NAD 1983 UTM Zone
coordinate system 13N
Geographic GCS North American 1
coordinate system 983
Data Format Source Type: Raster_ Discrete
Location /Access Colorado University at Denver
Date Server=; Service=esri_sde; Database=cuacel; User=sde; V ersion=SDE. DEFAULT 02/12/2008

Description: Common Vegetation Unit represents a combination of life form, species, and cover the study area of Mountain Plover
reclassswrega Type of data: Tabular digital data Clipped to Carrizo
P2 mp_regap.mdb Location /Access Date University Colorado at Denver WCEPHDNC2406- 2\D$\UserFiles\amit \GIS_Data\reclassswregap .mdb 02/16/08 Boundary, Reclassified the SWReGap vegetation land types on scale from 1 to 5, where 1 being least suitable and 5 being most suitable type of vegetation.

Description: The table represents a Common Vegetation Unit, including a combination of life form, species, and cover according to the study area of BTPD All land types unsuitable to Mountain Plover given a value of 0


Agnew W, Uresk DW, Hansen RM. 1986. Flora and fauna associated with prairie
dog colonies and adjacent ungrazed mixed-grass prairie in Western South
Dakota. Journal of Range Management 39(2): 135-139.
Augustine DJ, Cully JF, Johnson TL. 2007. Influence of fire on black-tailed
prairie dog colony expansion in shortgrass steppe. Rangeland Ecology and
Management 60(5):538-542.
Bahn V. 1998. Habitat requirements and Habitat Suitability Ilndex for the
Marbled Murrelet (Brachyramphus marmoratus) as a management target
species in the Ursus Valley, British Columbia. Marburg, Germany:
Phillips-Universita p. 167.
Bangert RK, Slobodchikoff CN. 2000. The Gunnison's prairie dog structures a
high desert grassland landscape as a keystone engineer. Journal of Arid
Environments 46:357-369.
Biggins DE, Kosoy MY. 2001. Influences of introduced plague on North
American mammals: implications from ecology of plague in Asia. Journal
of Mammalogy 82:906-916.
Bolstad P. 2005. GIS fundamentals: a first Txt on Geographic Information
Systems. White Bear Lake, Minnesota: Eider Press, p. 539.
Bonham CD, Lerwick A. 1976. Vegetation changes induced by prairie dogs on
shortgrass range. J. Range Manage 29(3):221-225.
Boulder Co. 2006. Black-tailed prairie dog component. Boulder CO.
Burger A, Bahn V. 2004. Inland habitat associations of Marbled Murrelets on
southwest Vancouver Island, British Columbia. Field Omithol 75(1):53-
CDOW. 2003. Conservation plan for grassland species in Colorado Denver,
Colorado: Colorado Division of Wildlife p. 210.

Ceballos G, Pacheco J, List R. 1999. Influence of prairie dogs (Cynomys
ludovicianus) on habitat heterogeneity and mammalian diversity in
Mexico. Journal of Arid Environments 41:161-172.
Clark TW, Campbell TM, Socha DG, Casey DE. 1982. Prairie dog colony
attributes and associated vertebrate species. Great Basin Naturalist 42:572-
Clippinger NW. 1989. Habitat suitability index models: black-tailed prairie dog.
U.S. Fish and Wildl. Serv. Report nr 82 (10.156). p. 21.
CNHP. 2006. Tracked mammal species.
Coppock DL, Detling JK, Ellis JE, Dyer MI. 1983. Plant-herbivore interactions in
a North American mixed-grass prairie. I. Effects of black-tailed prairie
dogs on intraseasonal aboveground plant biomass and nutrienit dynamics
and plant species diversity. Oecologia 56:1-9.
County B. 2002. Grassland management plan, prairie dog habitat element.
Developed for Boulder county parks and Open Space department.
Boulder, p. 93.
Crance JN. 1987. Guidelines for using the Delphi technique to develop habitat
suitability index curves U.S. Fish and Wildlife Service: Biological Report
Report nr 82(10.134). p. 21.
Crosetto M, Tarantola S. 2001. Uncertainty and sensitivity analysis: tools for GIS-
based model implementation. International Journal of Geographical
Information Science 15(5):415 437.
Cully J, J F, Williams ES. 2001. Interspecific comparisons of sylvatic plague in
prairie dogs. Journal of Mammalogy 82 894-905.
Delting JK. 1998. Mammalian herbivores: ecosystem-level effects in two
grassland national parks. Wildlife Society Bulletin 26:438-448.
Delting JK. 2006. Do prairie dog complete with livestock? In: Hoogland JL,
editor. Conservation of the black-tailed prairie dog Washington-Covelo-
London: Island Press, p. 65-88.

Donovan ML, Rabe DL, Olson CE. 1987. Use of Geographic Information
Systems to Develop Habitat Suitability Models. Wildlife Society Bulletin
ED AW I. 2000. Black-tailed prairie dog study of Eastern Colorado. Fort Collins,
p. 49.
EDOW. 2000. Black-tailed prairie dog study of Eastern Colorado. Fort Collins, p.
ESRI. 2000. Download Census 2000 TIGER/Line Data. ESRI.
ESRI. 2007. Understanding overlay analysis. ESRI.
Fahnestock JT, Larson DL, Plumb GE, Detling JK. 2003. Effects of ungulates and
prairie dogs on seed banks and vegetation in a North American mixed-
grass prairie. Plant Ecology 167(2):255-268.
Gillenwatera D, Granataa T, Zikab U. 2006. GIS-based modeling of spawning
habitat suitability for walleye in the Sandusky River, Ohio, and
implications for dam removal and river restoration. Eecological
engineering 28:311-323.
Gillihan SC, Hutchingsm SW. 1999. Best management practices for select bird
species of the Comanche National Grassland Brighton, Colorado.
Gribb WJ, Robinson SF, Hamerlinck JD. Applying ArcView Spatial Analyst
software's ModelBuilder to habitat conservation planning; 2001; San
Diego, California. Environmental Systems Research Institute.
Guenther D, Detling JC. 2003. Observations of cattle use of prairie dog towns
Rangeland Ecology & Management 56:410-417.
Gustafson EJ, Murphy DW, Crow TR. 2001. Using a GIS model to assess
terrestrial salamander response to alternative forest management plans.
Environmental Management 63:281-292.
Hazlett DL. 2004. Vascular plant species of the Comanche National Grassland in
southeastern Colorado. Fort Collins, CO: U.S. Department of Agriculture,

Forest Service, Rocky Mountain Research Station. Report nr RMRS-GTR-
130. p. 36.
Hoogland JL. 1995. The black-tailed prairie dog; social life of a burrowing
animal. Chicago, IL University of Chicago Press, p. 557.
Johnson L, Wyckoff J, Nuszdorfer F, Muller B. 2006. Colorado Statewide
Environmental Geodatabase. Prepared for Colorado department of
Transportation. Department of Civil Engineering, University of Colorado
Johnson WC, Collinge SK. 2004. Landscape effects on black-tailed prairie dog
colonies Biological Conservation 115:487-497.
Kotliar NB. 2000. Application of the new keystone-species concept to prairie
dogs: how well does it work? Conservation Biology 14(6): 1715-1721.
Kotliar NB, Baker BW, Whicker AD, Plumb G. 1999. A critical review of
assumptions about the prairie dog as a keystone species Environmental
Management 24:177-192.
Larson MA, Thompson FR, Millspaugh JJ, Dijak WD, Shifley SR. 2004. Linking
Population Viability, Habitat Suitability, and Landscape Simulation
models for conservation planning. Ecological Modeling 180:103-118.
Lauenroth WK, Milchunas DG, editors. 1992. Short-grass steppe. New York:
Elsevier.p. 183-226.
Lomolino MV, Smith GA. 2001. Dynamic biogeography of prairie dog (Cynomys
ludovicianus) towns near the edge of their range Journal of Mammalogy
Lomolino MV, Smith GA. 2003. Terrestrial vertebrate communities at black-
tailed prairie dog (Cynomys ludovicianus) towns. Biological Conservation
Lowry JH. 2005a. Southwest Regional Gap Analysis Project: final report on Land
Cover mapping methods, RS/GIS laboratory. Utah State University,
Logan, Utah. p. 50.
Lowry JH. 2005b. SWReGAP Landcover legend descriptions.

Lowry JH, Ramsey RD, Thomas K, Schrupp D, Sajwaj T, Kirby J, Waller E,
Schrader S, Falzarano S, Langs L et al. 2007. Mapping moderate-scale
land-cover over very large geographic areas within a collaborative
framework: A case study of the Southwest regional Gap Analysis Project
(SWReGAP). Remote Sensing of Environment 108:59-73.
Luce RJ. 2003. A multi-state conservation plan for the black-tailed prairie dog,
Cynomys ludovicianus, in the United States an addendum to the black-
tailed prairie dog conservation Assessment and Strategy, November 3,
Mackessy D, Stephen P. 1998. A survey of the herpetofauna of the Comanche
National Grasslands in Southeastern Colorado. In: Service UF, editor:
Jamestown, ND: Northern Prairie Wildlife Research Center Online, p. 61.
Malczewsk J. 2004. GIS-based Land-Use Suitability analysis: a critical overview.
Progress in Planning 62(l):3-65.
Marzluff JM, Millspaugh JJ, Ceder KR, Oliver CD, Withey J, McCarter JB,
Mason CL, Comnick J. 2002. Modeling changes in wildlife habitat and
Timber Revenues in response to forest management. Forest Science
48(2): 191-202.
Miller B, Biggins D, Wemmer C. 1990. A proposal to conserve black-footed
ferrets and the prairie dog ecosystem. Environmental Management
Miller B, Ceballos G, Reading R. 1994. The prairie dog and biotic diversity.
Conservation Biology 8(3):667-681.
Miller B, Reading R, Hoogland J, Clark T, Ceballos G, List R, Forrest S,
Hanebury L, Manzano K, Pacheco J et al. 2000. The role of prairie dogs as
a keystone species: response to Stapp. Conservation Biology 14(1):318-
Miller R, Knopf F. 1993. Growth and survivalof mountain plovers. Field
Ornithology 64(4):500-506.
Morrison ML, Marcot BG, Mannan RM. 2006. Wildlife-habitat relationships:
concepts and applications: Island Press.

Muller B, Johnson L, Wyckoff J, Nuszdorfer F, B. B. 2007. Areawide cumulative
effects analysis using GIS. 86th Annual Meeting of the Transportation
Research Board.
NatureServe. 2007. An online encyclopedia of life [web application] Arlington,
Virginia: NatureServe.
North M, Franklin J, Carey A, Forsman E, Hamer T. 1999. Forest stands structure
of the Northern Spotted Owls foraging habitat. Forest Science 45(4):520-
Orth PB, Kennedy PL. 2001. Do land-use patterns influence nest-site selection by
burrowing owls (Athene cunicularia hypugaea) in northeastern Colorado?
Canadian Journal of Zoology 79:1038-1045.
PDC. 2007. Black tailed prairie dog (Cynomys ludovicianus). Boulder: Prairie
Dog Coalition.
Proctor J, Haskins B. GIS model for identifying potential black-tailed prairie dog
habitat in the Northern Great Plains shortgrass prairie 1998; San Diego,
California. ESRI.
Proctor J, Haskins B, Forrest C. 2006. Focal area for conservation of prairie dogs
and the grassland ecosystem In: Hoogland JL, editor. Conservation of the
black-tailed prairie dog led. Washington Covelo London: Island Press,
p. 232-247.
Reading RP, Beissinger SR, Grensten JJ, Clark TW. 1989. Attributes of black-
tailed prairie dog colonies in northcentral Montana, with management
recommendations for the conservation of biodiversity. Montana BLM
Wildlife Tech. Bull. No. 2. .U.S. Department of the Interior, Bureau of
Land Management: 13-27. In cooperation with: Montana Department of
Fish, Wildlife, and Parks. Report nr 2.
Russell RE, Detling JK. 2003. Grasshoppers (Orthoptera: Acrididae) and black-
tailed prairie dogs (Sciuridae:Cynomys ludovicianus (Ord)): Associations
between two rangeland herbivores. Journal of Kansas Entomological
Society 76:578-587.

Ryke NC. 2005. Existing conditions and trend reports Cimarron and Comanche
National Grasslands (chapter23). USDA Forest Service.
Service NP. 2007. Black-tailed prairie dog management plan environmental
Shipley BK, Reading RP. 2006. A comparison of herpetofauna and small
mammal diversity on black-tailed prairie dog (Cynomys ludovicianus)
colonies and non-colonized grasslands in Colorado Journal of Arid
Environments 66(1):27-41.
Store R, Jokimaki R. 2003. A GIS-based Multi-Scale Approach to Habitat
Suitability Modeling. Ecological Modeling Ecological Modelling(169):l-
Store R, Kangas J. 2001. Integrating spatial multi-criteria evaluation and expert
knowledge for GIS-based habitat suitability modeling. Landscape Urban
Plan 55(2):79-93.
Tileston JV, Lechleitner RR. 1966. Some comparisons of the black-tailed and
white-tailed prairie dogs in north-central Colorado. American Midland
Naturalist 75:292-316.
Ulev E. 2007. Cynomys ludovicianus Fire Effects Information System (Online):
U.S. Department of Agriculture, Forest Service, Rocky Mountain
Research Station, Fire Sciences Laboratory .
USFWS. 1980. Habitat Evaluation Procedures (HEP). U.S. Fish and Wildl.
USFWS. 1981a. Standards for the development of Habitat Suitability Index
models for use in the Habitat Evaluation Procedures. ESM 103 ed: U.S.
Fish and Wildlife Service.
USFWS. 1981b. U.S. Fish and Wildlife Service. Standards for the development of
habitat suitability index models for use in the Habitat Evaluation
Procedures. Washington DC: U.S. Fish and Wildl. Service.
USFWS. 2004. Black-tailed prairie dog removed from candidate species list. U.S.
Fish and Wildlife Service.

Van Pelt WE. 1999. The black-tailed prairie dog conservation assessment and
strategy final draft. Nongame and Endangered Wildlife Program. In:
Arizona Game and Fish Dept. P, editor.
Whicker AD, Detling JK. 1998. Ecological consequences of prairie dog
disturbances. BioScience 38:778-784.
Winter SL, Cully F, Pontius JS. 2002. Vegetation of prairie dog colonies and non-
colonized short-grass prairie. Journal of range management 55:502-508.
Winter SL, Cully F, Pontius JS. 2003. Breeding season avifauna of prairie dog
colonies and non-colonized areas in shortgrass prairie. Transactions of the
Kansas Academy of Science 106(3/4): 129-138.
Wyckoff J, Johnson L, Nuszdorfer F. 2004. GIS- based cumulative effects
assessment. University of Colorado at Denver, GIS Program. Report nr
Wyckoff J, Nuszdorfer F, Johnson L, Muller B. 2007. Habitat Suitability Indices
for imperiled, critically imperiled, endangered and threatened species and
species of concern in Colorado. Denver: University of Colorado at Denver
and Health Sciences Center.