Citation
Raptor habitat relationships at the Rocky Mountain Arsenal National Wildlife Area

Material Information

Title:
Raptor habitat relationships at the Rocky Mountain Arsenal National Wildlife Area
Creator:
Beane, Ronald D
Publication Date:
Language:
English
Physical Description:
ix, 101 leaves : illustrations ; 28 cm

Subjects

Subjects / Keywords:
Birds of prey -- Colorado ( lcsh )
Birds of prey -- Ecology ( lcsh )
Habitat selection ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaves 94-101).
General Note:
Submitted in partial fulfillment of the requirements for the degree, Master of Arts, Biology.
Statement of Responsibility:
by Ronald D. Beane.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
36434441 ( OCLC )
ocm36434441
Classification:
LD1190.L45 1996m .B43 ( lcc )

Downloads

This item has the following downloads:


Full Text
RAPTOR HABITAT RELATIONSHIPS AT THE ROCKY MOUNTAIN ARSENAL
NATIONAL WILDLIFE AREA
by
Ronald D. Beane
B.S., Colorado State University, 1980
A thesis submitted to the
University of Colorado at Denver
in partial fulfillment
of the requirements for the degree of
Master of Arts
Biology
1996
AL


This thesis for the Master of Arts
degree by
Ronald D. Beane
Diana F. Tomback
Jim Koehler
Douglas P. Reagan
11


Beane, Ronald D. (M.A. Biology)
Raptor Habitat Relationships at the Rocky Mountain Arsenal National Wildlife Area
Thesis directed by Dr. Charles R. Preston
ABSTRACT
I investigated habitat use and distribution of diurnal raptors during a three years
study at the Rocky Mountain Arsenal National Wildlife Area (RMA), northeast of
Denver, Colorado. I used roadside counts to document seasonal relative abundance
and distribution of diurnal raptors and radio-tracking studies to determine habitat use
and general winter ecology of overwintering ferruginous hawks. Fourteen diurnal
raptor species were identified at RMA. The most frequently encountered species were
American kestrel (Falco sparverius), Swainsons hawk (buteo swainsoni) and red-
tailed hawk (Buteo jamaicensis) in summer, and bald eagle (Haliaeetus
leucocephalus), ferruginous (Buteo regalis), red-tailed, and rough-legged hawk (Buteo
lagopus) in winter. The mean number of raptors per summer survey was 21.0 in 1991
and 20.3 in 1992. Mean number of winter raptors observed increased from 24.0 in
1991-1992 to 30.7 in 1993-1994, due to increased numbers of ferruginous hawk. The
increase in ferruginous hawks was likely in response to increased prairie dog
populations. Winter distribution patterns confirmed that red-tailed hawks are
associated with more trees than sympatric hawks at RMA, and ferruginous hawks are
in


distributed in response to a prairie dogs. Diurnal raptors did at RMA not use habitat
in proportion to availability. Weedy forb areas were preferred by summer American
kestrels and winter ferruginous hawks. Wintering bald eagles, rough-legged hawks and
red-tailed hawks preferred wooded/riparian and/or wetland habitats. Habitats avoided
by most diurnal raptors in both summer and winter included crested wheatgrass and
cropland. Ordination of species along habitat gradients (PCA) showed that red-tailed
and Swainsons hawks occured in habitats with relatively lower diversity and more
perches, whereas American kestrels occurred in relatively diverse habitats with few
trees and perches. The presence of prairie dogs was the overriding factor in
determining ferruginous hawk habitat use. All habitats with prairie dogs were
preferred, and most habitats without prairie dogs were avoided. Winter home ranges
of ferruginous hawks became smaller in each successive year of the study ( 29.5 km in
1991-1992, 14.4 km2 in 1992-1993, and 8.3 km2 in 1993-1994). This reduction in
home range size coincided with increased abundance of prairie dogs. Overwintering
ferruginous hawks showed strong fidelity to communal roosts.
This abstract accurately represents the content of the candidates thesis. I reccommend
its publication
Charles R. Preston
IV


V


DEDICATION
I would like to dedicate this work to my loving wife, Janet, and my children, Chad and
Kara who withstood countless days and hours without a husband and father to provide
them with guidance, comfort and encouragement. Their patience and understanding is
incredible.
vi


CONTENTS
Acknowledgments.............................................ix
CHAPTER
1. OVERVIEW.............................................1
Study Area........................................5
2. RAPTOR ABUNDANCE AND DISTRIBUTION AT ROCKY
MOUNTAIN ARSENAL NATIONAL WILDLIFE AREA
Introduction...................................................8
Methods........................................................9
Results.......................................................13
Summer.....................................................13
Winter.....................................................18
Migration..................................................22
Discussion....................................................23
3. HABITAT USE OF DIURNAL RAPTORS AT ROCKY
MOUNTAIN ARSENAL NATIONAL WILDLIFE AREA
Introduction.......................................................34
Methods............................................................35
Habitat Classification.........................................35
Roadside Counts................................................38
Principal Component Analysis...................................39
Results............................................................41
Results of Habitat Use and Availability fro Roadside counts....41
Principal Component Analysis of Roadside Raptor Counts.........48
vu


Discussion................................................59
4. WINTER ECOLOGY OF FERRUGINOUS HAWKS AT ROCKY
MOUNTAIN ARSENAL NATIONAL WILDLIFE AREA
Introduction..............................................63
Methods...................................................64
Habitat Use............................................64
Home Range of Wintering Ferruginous Hawks..............66
Roost Site Characteristics.............................67
Results...................................................68
Analysis of Habitat Use from Radio-tracking............68
Analysis of Habitat Use from Roadside Counts...........70
Home Range Results.....................................74
Description of Communal Roosts.........................79
Discussion................................................83
Habitat Use............................................83
Home Range.............................................86
Communal Roosts........................................88
5. OVERALL CONCLUSIONS AND DISCUSSION.......................91
REFERENCES CITED.......................................................94
viii


ACKNOWLEDGMENTS
Funding for this project was provided by the US Army, under a cooperative
agreement among the Army, US Fish and Wildlife Service (USFWS), and the Denver
Museum of Natural History (DMNH). I would like to express deep gratitude to my
graduate committee members, Dr. D Tomback, Dr. J. Koehler, Dr. D. Reagan and in
particular my graduate advisor, Dr. C. Preston ,who provided incomparable guidance,
patience and encouragement. I wish to thank the entire staff of the Zoology Department
at the DMNH Dr. R. Pieglar, Dr. C. Jones, Dr. C. Meaney, and M. Castenenda.
Personnel from the USFWS, especially D. Gober, L. Malone, M. Lockhart, J.Griess,
and D. Matiatos provided field assistance, helpful suggestions and logistical support.
D. Bockus of the USFWS conducted most of the diet analysis and provided invaluable
field assistance. I am also thankful to C. Mackey, Morrison Knudsen Environmental
Services, Inc. for providing insight into vegetation communities at RMA, and to my
field assistants, particularly R. DeBaca and F. Hein who made data collection efficient,
rewarding and fun.
IX


CHAPTER 1
OVERVIEW
In 1991 the Denver Museum of Natural History (DMNH) began a series of
ecological studies at the Rocky Mountain Arsenal National Wildlife Area (RMA)
(Figure 1.1). These investigations evaluated the habitat associations of small birds,
small mammals, lagomorphs and raptors. The Rocky Mountain Arsenal National
Wildlife Area is currently on the National Priorities List as a 'Superfund Site' and is
undergoing environmental clean-up by the Army and Shell Oil Company (USFWS
1996). Current interim clean-up actions and eventual remediation of hazardous
waste associated with RMA could potentially impact thousands of hectares of
wildlife habitat. Wildlife management decisions and mitigation of environmental
impacts of remediation will be implemented based on the best available information.
Therefore, it is imperative to understand wildlife-habitat relationships, including
raptor-habitat relationships, prior to remediation.
Diurnal raptors are the most visible predators at RMA. Common diurnal
raptors at RMA are red-tailed hawks (Buteo jamaicensis), Swainsons hawks (Buteo
swainsoni), northern harrier (Circus cyaneus) and American kestrels (Falco
sparverius) in summer; ferruginous hawks (Buteo regalis), red-tailed hawks, rough-
legged hawks (Buteo lagopus), bald eagles (Haliaeetus leucocephalus) and golden
1


2


eagles in winter (Aquila chrysaetos). Because of their position at the top of the food
chain and their great mobility, raptors are valuable as environmental barometers
(Craighead 1987). Diurnal raptors also control populations of herbivores (e.g.,
rodents, grasshoppers, rabbits and hares) and are an important component of
grassland ecosystems.
In 1987 the National Wildlife Federation sponsored a series of raptor
management symposia held regionally across the United States to discuss the status
and importance of raptors. Both the Western and Southwestern Raptor Management
Symposia identified winter habitat requirements of ferruginous hawks as an important
future research need (LeFranc and Glinski 1988, Harlow and Bloom 1989). The
ferruginous hawk (Buteo regalis) is a frequent overwintering raptor at the Arsenal.
This species is endemic to grassland and open shrubland habitats in North America,
and is thought to be declining throughout much of its breeding range (Olendorff
1973, Houston and Bechard 1984, Schmutz 1984); however, declines have only been
confirmed in Utah and Nevada (Bechard and Schmutz 1995). Because of the high
visibility and ecological importance of diurnal raptors on RMA, the U.S. Fish and
Wildlife Service (USFWS) felt additional information was needed on raptor habitat
relationships, particularly habitat relationships of ferruginous hawk.
The purpose of this study was to examine the temporal and spatial use of
habitats by all common diurnal raptors and particularly to describe winter habitat
requirements of ferruginous hawks.
3


My objectives were to:
1. Document the year-round relative abundance and distribution of common
falconiforms at RMA.
2. Determine seasonal habitat use of common diurnal raptors at RMA.
3. Determine winter habitat use of ferruginous hawks at RMA.
4. Document average home range of ferruginous hawks overwintering at RMA.
5. Identify and characterize ferruginous hawk communal roost sites on RMA.
I met my objectives with a combination of descriptive studies and focused
hypothesis testing. The primary focus of my study was to test the null hypotheses
that 1) raptors use summer and winter habitat in direct proportion to its availability;
and specifically, 2) wintering ferruginous hawks use habitat in direct proportion to its
availability.
I used ancillary descriptive studies to document relative abundance and
distribution of diurnal raptors in general, and to specifically describe average home
range and communal roost site characteristics of overwintering ferruginous hawks.
My study is divided into five separate chapters. This first chapter provides a general
introduction, study organization, and study area description. The second chapter
describes the relative abundance and distribution or diurnal raptors at RMA. The
third chapter builds on Chapter 2, describes seasonal habitat use, and develops an
ordination of coexisting diurnal raptors with respect to habitat components at RMA.
4


The forth chapter focuses on the winter ecology of ferruginous hawks and builds on
the general habitat use patterns described in Chapter 3. Chapter 5 provides overall
summary and conclusions of my study. Detailed descriptions of methods used to
meet my study objectives are provided in each chapter.
Study Area
My study area consisted of the 6800 ha Rocky Mountain Arsenal located 16
km northeast of downtown Denver, Colorado (Figure 1.1). Originally shortgrass
steppe dominated by blue grama (Bouteloua gracilis) and buffalograss (Buchloe
dactyloides) (Lauenroth 1991), the RMA was converted to agriculture in the early
1900s. The area was purchased in 1942 by the U.S. Army for the production of
chemical weapons for World War II. After the war, from 1947 to 1982, the Army,
Shell Chemical and other chemical contractors manufactured chemical weapons and
commercial pesticides at the installation. The area was listed as a federal Superfund
site in the mid-1980s under the Comprehensive Environmental Reclamation
Conservation and Liabilities Act (CERCLA) and is currently undergoing cleanup by
the Army and Shell Oil Company. The Rocky Mountain Arsenal National Wildlife
Refuge Act of 1992 (Public Law 102-402) established the RMA as a National Wildlife
Refuge, pending environmental cleanup. Until then RMA is considered a National
Wildlife Area.
5


RMA is characterized by open expanses of disturbed and native grasslands.
Disturbed grasslands are dominated by cheatgrass (Anisantha tectornm), kochia
(Bassia sieversiana), field bindweed (Convolvulus arvensis), and planted areas of
crested wheatgrass (Agropyron cristatum). Native grasslands are dominated by sand
dropseed (Sporobolus cryptandrus), needle-and-thread grass (Stipa comat a), and red
threeawn (Aristida longiseta). Scattered among the open grasslands are shrub-
grassland areas dominated by yucca (Yucca glauca) and sand sagebrush (Artemisia
fdifolia). The southern portion of my study area is characterized by grassland
interspersed with woodland and riparian areas consisting of plains cottonwood
(Populas sargentii), white poplar (Populas alba), locust (Robina spp.), and Chinese
elm (Ulmuspumila). Three reservoirs (38, 25 and 3.2 ha, respectively) and First Creek,
an intermittent stream that flows from south to north through RMA, are the only
reliable water sources within 10 km of the study area. The area surrounding RMA
consists of cultivated cropland on the north and east, and urban development, including
Stapleton International Airport, on the south and west. Although surrounded by urban
development, RMA remains a large island of open prairie habitat that supports diverse
and abundant prey populations
Currently, the RMA is experiencing various degrees of habitat alteration as a
result of interim response cleanup actions. These disturbances are anticipated to
6


increase in the near future and continue for several decades as final remediation actions
are implemented.
7


CHAPTER 2
RAPTOR ABUNDANCE AND DISTRIBUTION AT ROCKY MOUNTAIN
ARSENAL NATIONAL WILDLIFE AREA
Introduction
Little is known about the distribution and abundance of raptors historically
within the shortgrass steppe near Denver, Colorado, particularly at RMA, which was
closed to the public for decades. With the establishment of RMA as a National
Wildlife Area, information was needed on wildlife habitat requirements to effectively
manage the resource. Additionally, information on fauna at RMA, particularly top
carnivores, was needed to establish a pre-remediation baseline. A study of the short-
term relative abundance and distribution of raptors at RMA can help determine
trends of raptor occurrence within the boundaries of RMA and to some extent within
the general region. This information can provide resource managers with
information to identify and manage high use and low use areas on a species-specific
basis.
My objective for this study was to document the distribution and relative
abundance of common diurnal raptors at RMA.
8


Methods
In order to determine relative abundance and distribution I conducted sixty
roadside counts for all diurnal raptors for two consecutive years. These roadside
counts were conducted semimonthly from June 1991 through June 1993, following
the methods of Fuller and Mosher (1981, 1987). I conducted additional roadside
counts in summer and fall 1993 to correspond with companion lagomorph studies.
I conducted more counts in winter because the primary focus of this investigation
was the distribution and relative abundance of wintering ferruginous hawks. I
conducted roadside counts between 06:00 and 11:00 hours on days with favorable
conditions for raptor observation (visibility exceeding 1500 meters and wind velocity
less than 16 km/hour). My observations were conducted from a vehicle moving 25
to 40 km/hour, and all raptors observed within 400 meters of the road were recorded.
During roadside counts I recorded start and finish odometer readings, start and finish
times, location of bird (odometer reading, Township/Range section, and map plot)
distance to the bird (calculated from a map showing measured distances to perch
sites), vegetation type surrounding the bird, proximity to human disturbance, bird
behavior, and perch type. Data for all years were pooled by season for the analysis
of raptor distribution.
9


My raptor roadside count route was 39 km long, consisting of maintained
section roads within the boundaries of RMA (Figure 2.1). This route was
subdivided into a series of 24 sample segments of approximately equal length (1.6
km). Starting and stopping points for each segment corresponded to section roads.
In order to maintain independence of observations, overlapping portions of sample
segments were eliminated. Each segment was treated as an independent sample plot
with specific habitat and land use characteristics.
I divided the study area into five ecologically similar regions that were
representative of the vegetation, topography, land use, soils and human presence
within the study area (Figure 2.2). Regions consisted of 1) the northwestern region,
characterized by native grasses, cheatgrass, and crested wheatgrass; 2) the
southwestern region, characterized by scattered pockets of trees and windbreaks,
native grasses, crested wheatgrass, and frequent human presence (passenger and
private aircraft, heavy vehicle traffic, administrative and maintenance buildings); 3)
the southeastern region, characterized by a mosaic of lakes, canals, wetlands, wooded
riparian areas, and shrublands; 4) the northeastern region characterized by open
native grassland, prairie dog colonies and a narrow riparian corridor along First
Creek, and; 5) the central region, characterized by historic and current human
disturbance such as industrial and chemical facilities, administrative buildings and
areas of chemical contamination.
10


9th AVE
8th AVE
7th AVE
Figure 2.1 Raptor roadside count transect route
at Rocky Mountain Arsenal NWA
Legend
Established section road
Transect route
11


Figure 2.2 Regions used for evaluation of raptor distribution and abundance
on Rocky Mountain Arsenal NWA
12


I divided counts into summer (breeding season), winter, and fall and spring
migration. Following the methods of Fuller and Mosher (1987) I described species
abundance and distribution within the RMA by recording the number of diurnal
raptors observed within ecologically similar regions.
I evaluated distribution and abundance within each region for the summer
breeding season (May through August) and winter (November 8 through February)
for all raptors that averaged more than one observation per roadside count. Because
the five regions varied in size, I standardized results by comparing the number of
diurnal raptors observed per km of survey route within each region.
Results
Summer
Eleven summer counts were conducted between 15 June 1991 and September
1992. Three species, red-tailed hawk, Swainson's hawk, and American kestrel were
observed nesting at RMA. Additionally, non-breeding ferruginous hawks, northern
harriers and golden eagles were occasionally observed in summer (Table 2.1). The
average total number of raptors observed during summer roadside counts was 21.0
(SD = 2.35, n = 5) in 1991 and 20.3 (SD = 6.71, n = 6) in 1992. The American
kestrel was the most abundant species observed in each summer (Figure 2.3),
reaching peak abundance in late June and early July coinciding with fledging of
13


Table 2.1 Diurnal raptor species occurrence by season at
Rocky Mountain Arsenal National Wildlife Area.
Species Season
Winter Spring Summer Fall
Red-tailed hawk X X X X
Swainson's hawk X X X
Ferruginous hawk X X X X
Rough-legged hawk X X X
Cooper's hawk X
Northern harrier X X X X
Bald eagle X X X
Golden eagle X X X X
American kestrel X X X X
Peregrine falcon X X X
Prairie falcon X X X
Merlin X
Osprey X X
Turkey vulture X X
14


1991
Figure 2.3 Mean number of raptors by species observed on roadside counts conducted
on Rocky Mountain Arsenal NWA Summer
Legend: FH Ferruginous Hawk
RTH Red-tailed Hawk
SH Swainsons Hawk
NH Northern Harrier
AK American Kestrel
GE Golden Eagle
15


young. Both red-tailed hawks and Swainson's hawks reached peak abundance during
August in both years as young fledged and became more visible.
Analysis of summer distribution of raptors at RMA by region revealed
patterns for each of the three primary breeding raptors. American kestrels were most
numerous in central and northeastern regions (Figure 2.4a). These regions were
characterized by open landscapes dominated by weedy forbs, and the central region
contained some of the most chemically polluted and physically disturbed areas on
RMA. Distribution of American kestrels varied somewhat between 1991 and 1992.
In 1992 the average number of kestrels per km observed per transect decreased by 50
percent in the northeastern region and doubled in the northwestern and
southwestern regions (Figure 2.4a).
Red-tailed hawks were unevenly distributed over RMA, occurring most often
in the central region, and rarely in the southeastern region (Figure 2.4b).
Comparison of data between years showed fewer red-tailed hawks were observed in
1992.
The number of Swainson's hawks was also unevenly distributed across
regions of RMA. The greatest abundance occurred in the southwestern region
(Figure 2.4c). I rarely observed Swainsons hawks in the southeastern region.
Ferruginous hawks occurred sporadically in summer at RMA as vagrants or
non-breeding subadults and were not observed to breed at the study area. I generally
16


6.00
5.00
4.00
3.00
2.00
1.00
0.00
B1991 AK
1992 AK
NW SW SE Central NE
a. American Kestrel
b. Red-tailed Hawk
0.7
NW SW SE Central NE
c. Swainson's Hawk
Figure 2.4 Mean number of summer diurnal raptors per km observed during roadside counts
on Rocky Mountain Arsenal NWA. Data are presented by region and year
17


observed ferruginous hawks in northern regions in summer, especially the
northeastern region. However, the mean number of observations of ferruginous
hawks was less than one per roadside count and insufficient for further evaluation.
Winter
I conducted twenty-five winter roadside counts between 15 November 1991
and 30 January 1994. Winter counts confirmed that the ferruginous hawk, red-tailed
hawk, rough-legged hawk (Buteo lagopus), and bald eagle (Haliaeetus
leucocephalus), were common winter residents at RMA. Other species observed
included golden eagle, northern harrier, American kestrel, merlin (Falco
columbarius), prairie falcon, (Falco mexicanus), Coopers hawk (Accipiter cooperii)
and turkey vulture (Cathartes aura) (Table 2.1). The average number of total
raptors observed per winter count increased from 24.0 (SD = 7.3, n = 11) in 1991-92
to 30.7 (SD = 6.5, n = 6) in 1993-94, primarily due to a notable increase in the
average number of ferruginous hawks. Ferruginous hawks increased from an
average of 8.6 (SD = 2.7, n = 11) hawks per count in 1991-92 to 11.5 (SD = 4.8, n =
8) and 12.2 (SD = 2.5, n = 6) hawks per count in 1992-93 and 1993-94, respectively.
The average number of red-tailed hawks observed was consistent over the three
winters (5.9, 6.6, and 5.8); bald eagles and rough-legged hawks varied slightly
from year-to-year; golden eagles showed a steady increase over the three winters;
18


and northern harriers increased slightly in 1993-94 (Figure 2.5). Ferruginous hawk
occurrence was greatest in the northeastern, southwestern and central regions, with
few hawks observed in the southeastern region (Figure 2.6a). Regional distribution
was similar in all years, but a slight decrease in ferruginous hawk abundance in
southern regions was observed in 1993-94. This corresponded with an increase in
abundance in the northwestern region (Figure 2.6a).
Winter distribution of red-tailed hawks was fairly consistent over the three
winters of observations. These birds occurred most frequently in the southwestern
and southeastern regions (Figure 2.6b).
Rough-legged hawks occurred most frequently in the central, southeastern
and northeastern regions (Figure 2.6c). Distribution of rough-legged hawks on a
yearly basis was fairly consistent for the three winters of observations except that the
average number of rough-legged hawks observed per roadside count doubled in the
central region and decreased by 50 % in the southeastern region in 1992 (Figure
2.6c).
The distributions of both bald eagles and golden eagles were variable from
year to year in winter at RMA. I occasionally observed golden eagles during
roadside counts, primarily in the northeastern region. However, counts of golden
eagles averaged less than one per roadside count and were not further evaluated.
Bald eagles occurred in the southeastern and northeastern regions and generally
19


POOLED
FH
BIRTH
RLH
NH
AK
BE
HUGE
n fh
RTH
RLH
NH
AK
BE
1GE
Figure 2.5 Mean number of raptors by species observed on roadside counts conducted
on Rocky Mountain Arsenal NWA Winter
Legend: FH Ferruginous Hawk AK American Kestrel NH Northern Harrier
RTH- Red-tailed Hawk BE Bald Eagle
RLH Rough-legged Hawk GE Golden Eagle
20


0.50
0.40
0.30
0.20
0.10
0.00
JI
NW SW
a. Ferruginous Hawks
SE
Central
01991-92 FH
1992-93 FH
1993-94 FH
0.50
0.40
0.30
0.20
0.10
0.00

nr 1991-92 RTH 1992-93 RTH 1993-94 RTH

NW SW
b. Red-tailed Hawks
SE
Central
NE
0.30
0.25
0.20
0.15
0.10
0.05
0.00
NW SW
c. Rough-legged Hawks
SE
Central
1991-92 RLH
1992-93 RLH
1993-94 RLH
1991-92 BE
1992-93 BE
1993-94 BE
NW SW
d. Bald Eagles
Central
Figure 2.6 Mean number of wintering diurnal raptors per km observed during roadside count
counts on Rocky Mountain Arsenal NWA. Data are presented by region and year
21


avoided central and northwestern regions (Figure 2.6d). I most often observed bald
eagles in structurally similar habitat consisting of large mature cottonwoods with
open canopies and horizontal branches for perching. The southeastern and
northeastern regions also contained bald eagle manipulation stations (artificial
feeding stations) that tended to attract bald eagles to specific areas and accounts for
some of the distribution patterns observed.
Migration
Twelve raptor roadside counts were conducted during fall migration periods
(1 September 8 November); seven in 1991 and five in 1992. Twelve roadside
counts were also conducted during spring migration periods (1 March 30 April);
six each in 1992 and 1993. The average number of raptors per fall transect was
similar in 1991 (28.3, SD = 7.0, n = 7) and 1992 (29.2, SD = 6.7, n= 5). Spring road
transect counts showed a marked increase in the average number of raptors per
transect in 1993 (24.3, SD = 9.4, n = 6) over 1992 (28.8, SD = 8.5, n= 6).
Observations of raptors during migration periods are highly variable and are
dependent on weather patterns, prevailing winds, and the number of nonmigratory
raptors present. This variability was apparent during roadside counts of several
raptor species conducted during migration periods. The red-tailed hawk was the
most frequently encountered raptor observed during fall migration in both years,
22


accounting for 36.5 % of observations in 1991 and 48.0 % of observations in 1992.
The ferruginous hawk was the next most prevalent diurnal raptor during fall
migration with about 21 % of observations in both years, followed by Swainsons
hawk with 16.8 and 9.8 of observations in 1991 and 1992. Red-tailed hawks per
transect nearly doubled in spring 1993 (8.83, SD = 2.9, n = 6) from spring 1992
(4.83, SD = 4.8, n = 6) counts. American kestrels migrated through the study area in
large numbers in spring 1992, accounting for nearly 50 % of observations. However,
few kestrels occurred in spring 1993. Northern harriers were infrequently observed
during both fall and spring migration periods. However, harriers showed a big
increase in average number per transect in spring 1993. Other species occasionally
migrating through the study area included ospreys (Pandion haliaetus). peregrine
falcons (Falco peregrinus), prairie falcons and golden eagles.
Discussion
Frequently observed diurnal raptors at RMA included red-tailed hawks,
Swainsons hawks and American kestrels. American kestrel was the most frequently
observed species. Frequently observed wintering raptors at RMA included
ferruginous hawks, red-tailed hawks, rough-legged hawks and bald eagles. The
ferruginous hawk increased in abundance each year of the study, while the
abundance of red-tailed hawk, rough-legged hawk and bald eagle was relatively
23


stable. The red-tailed hawk is the only species of raptor at RMA that is relatively
abundant in both winter and summer.
Detectability varies among raptor species according to body size, seasonal
behavior, habitat use and flight behavior (Fuller and Mosher 1981, 1987). This
difference in detectability may have biased relative abundance estimates for diurnal
raptors at RMA, particularly for breeding American kestrels. Observations of
kestrels likely decreased as distance to the road increased due to their small body size
(Millsap and LeFranc 1988). Red-tailed and Swainsons hawks are relatively similar
in size and it is assumed they are equally detectable in RMA habitats. The three
major wintering buteos at RMA (ferruginous, red-tailed and rough-legged hawks) are
also relatively similar in body size and winter behavior and are assumed to be
equally detectable in winter. I also minimized differences in detectability by
standardizing the time of day, seasons and weather of roadside counts (Fuller and
Mosher 1987).
Red-tailed hawks were most often observed during the breeding season in
central regions containing scattered groves of large cottonwoods for perching and
nesting, and disturbed areas containing a high proportion of sparse vegetation and
bare ground that may increase prey availability. Disturbed areas near waste basins
and abandoned industrial complexes on RMA support a higher abundance of small
mammals, particularly deer mice (Peromyscus maniculatus) than more native
24


shortgrass habitats (MKE 1989, Boone and Preston 1994). I rarely observed red-
tailed hawks in the southeastern region, although this region contained abundant
riparian vegetation along lakes and canals. However, this region is also subject to
fishing and wildlife viewing pressure along the road transect route. Red-tailed hawks
may occur in areas further removed from human disturbance beyond the 400m
transect width and thus be more abundant in the southeastern region than indicated
by road transects. Although the typical width of roadside transects is 400 to 800 m
wide (Kochert 1986, Millsap and LeFranc 1988), Millsap and LeFranc (1988) found
that detectability of raptors decreased with increased distance and vegetation density.
These researchers showed that detectability in summer deciduous woodland
decreased sharply for models of three woodland hawks, including red-tailed hawk.
Based on these considerations, some red-tailed hawks were undoubtedly undetected
during summer roadside counts. However, detectability is significantly higher in
grassland than other vegetation types and consistent up to 300 m (Millsap and
LeFranc 1988). These authors also suggested adjustment of counts to account for
detectability will improve accuracy but lower precision. Furthermore, unadjusted
counts might be superior for monitoring studies where trends in raptor numbers over
time is the objective (Millsap and LeFranc 1988). Another consideration is the
general low abundance of summer red-tailed hawks at RMA. Typically, only one to
25


three pairs of red-tailed hawks breed at RMA (ESE 1989, USFWS 1995) and a
single nest near the roadside transect could influence observations and bias results.
Ferruginous hawks currently do not breed at RMA and no historical accounts
of breeding exist, although an occasional ferruginous hawk is observed in northern
regions of RMA during summer. These northern regions are characterized by open
shortgrass steppe and weedy forb habitats with large concentrations of prairie dogs.
The total number of ferruginous hawks observed was low (less than one observation
per transect), and distribution could be heavily influenced by a few individual hawks.
The USFWS has installed artificial nest structures at RMA, but they have not
attracted any breeding ferruginous hawks.
The distribution of American kestrels was somewhat variable between years
with a decrease of kestrels in the northeastern region coinciding with an increase in
the southwestern and northwestern region. However, American kestrels in all
regions appeared to be attracted to weedy areas. These areas typically support large
concentrations of grasshoppers and other insects at RMA (Beane, Unpubl. Data) that
provide prey for American kestrels (Sherrod 1978). Shifts in the summer
distribution of breeding raptors could be caused by the relocation, abandonment, or
establishment of one or more nest sites, or be a reflection of individual nest site
success. Any successful nest site within the roadside count transect would increase
the count within that specific region. An example of the influence of nest sites on
26


summer distribution patterns of raptors was evident in the variability of kestrel
distribution between years and the notable decrease of Swainson's hawks in the
southwestern region of RMA in 1992. Two Swainsons hawk nest sites occurred
within or near the transect corridor in this region in 1992, and the decreased
abundance recorded during roadside counts probably resulted from decreased nest
and fledging success between years.
The diversity and abundance of wintering diurnal raptors at RMA indicate
that the area is regionally important to wintering raptors. RMA supports one of the
largest communal bald eagle roosts within the South Platte River Basin.
Additionally, Christmas urban bird counts conducted by the Denver Audubon
Society indicate that RMA supports more rough-legged hawks, bald eagles, and
golden eagles than other areas in the Denver-metro region (Kingery 1996).
The increase in wintering ferruginous hawks observed during road transects
coincides with increased density of prairie dog populations (USFWS 1994) and
likely is a result of the expanded prey base attracting hawks to RMA rather than a
trend in the wintering population of ferruginous hawks in the region. A review of
urban Denver Christmas bird count data from 1988 through 1995 showed a general
decline in the number of ferruginous hawks observed within the Denver-metro
region, including the RMA. Some of this decline can be attributed to fluctuating
prey populations. Prairie dogs at RMA were severely reduced by an outbreak of
27


sylvatic plague in 1988-89, and wintering ferruginous hawk populations declined
accordingly. Some of the regional decline in 1989-90 was offset by healthy
populations of prairie dogs and wintering ferruginous hawks at Buckley Air National
Guard Base (ANGB) (Beane Unpubl. Data, Kingery 1996). Prairie dogs at Buckley
ANGB experienced sylvatic plague in 1989, and regional ferruginous populations
further declined (Kingery 1996). A recovery of prairie dogs at RMA through 1994
was reflected in a steady increase in populations of wintering ferruginous hawks,
both at RMA and regionally (USFWS 1994, Kingery 1996). A second outbreak of
plague at RMA in 1994 and continued low populations of prairie dogs at Buckley
ANGB, combined with unprecedented growth along the Colorado front range that
eliminated numerous prairie dog colonies, greatly reduced counts of wintering
ferruginous hawks in the Denver area.
Wintering red-tailed hawks were associated with riparian habitat along First
Creek in the southeastern and northeastern regions and woodland and shelterbelts in
the southwestern region, as well as abundant prairie dogs (Cynomys ludovicianus)
and cottontail rabbits (Sylvilagus auduboni) for prey. These observation are
consistent with studies that found red-tailed hawks associate more with trees than do
sympatric ferruginous and rough-legged hawks (Schnell 1968, Janes 1985).
Researchers studying habitat partitioning between breeding red-tailed, ferruginous
and Swainsons hawks have found that the three species exhibit a high degree of
28


dietary and habitat overlap, and segregate primarily by partitioning nesting habitat
(Cotrell 1981, Restani 1991)
A comparison of summer and winter distribution of red-tails at RMA
revealed a seasonal shift away from southern regions in winter to a more central
distribution in summer (Figure 2.7). This could be the result of seasonal changes in
prey availability, the availability of nest sites, interactions with other nesting raptors,
or differences in detectability of raptors in wooded areas. However, most wooded
areas at RMA consist of relatively open and dispersed clumps of trees that permit
good visibility within 400m of the road transect.
Human presence and activity are known to alter raptor behavior (Stalmaster
and Newman 1978, Knight and Knight 1984, Andersen et al. 1986, 1989, 1990,
Holmes et al 1993). In this study diurnal raptors wintering at RMA did not avoid
areas of heavy human pedestrian traffic in the central region of the arsenal. These
results are consistent with Preston and Beane (1996), who, using a different analysis
(logit) found that raptors at RMA did not avoid heavily used roadways and may have
habituated to relatively high levels of human activity found at RMA.
A comparison of buteos wintering at RMA revealed several interesting
distribution patterns (Figure 2.8). Bald eagles were not included in this
comparison because the practice of establishing eagle feeding stations may bias their
distribution. Equal numbers of red-tailed hawks and ferruginous hawks were
29


0.35
NW SW SE Central NE
Figure 2.7 Comparison of summer and winter distribution of red-tailed hawks
on Rocky Mountain Arsenal NWA 1991-1994
Legend: RTH Red-tailed Hawk
30


Figure 2.8 Mean number of wintering buteos per km observed during roadside counts
on Rocky Mountain Arsenal NWA
Legend: FH Ferruginous Hawk
RTH Red-tailed Hawk
RLH Rough-legged Hawk
31


observed in the southwestern region which accounted for this region supporting the
highest number of buteos per km at RMA. This region has also been identified as
land to be auctioned or converted to commercial/industrial uses as part of the RMA
National Wildlife Area Act of 1992 (H.R. 1435). Two regions that also supported a
high abundance of ferruginous hawks (central and northeastern) had relatively low
abundance of red-tailed hawks. Conversely, the southeastern region supported a
high relative abundance of red-tailed hawks per km and the lowest number of
ferruginous hawks per km at RMA.
Few buteos occurred in the northwestern region over the three years of
observations. This region consists of large areas of disturbed and native prairie and
vast expanses of crested wheatgrass that has good potential for habitat enhancement
for raptors. Some specific enhancement could include managing prey populations,
particularly prairie dogs; providing habitat structure for alternate prey, such as
rabbits and mice; removing nesting, perching and foraging habitat from areas of
contamination; providing poles and structures for perching, and trees and wind
breaks for roosting and nesting in areas distant to contamination; providing visual
shields to screen centers of human activity; managing sylvatic plague, and;
coordinating with local, county, and state governments to preserve and protect buffer
areas and surrounding open space to maintain the ecological function of the RMA.
Without buffer zones, other nearby open space, and movement corridors to connect
32


RMA to surrounding habitat, many of the summer and winter raptor populations
would substantially decrease in number or disappear altogether.
33


CHAPTER 3
HABITAT USE OF DIURNAL RAPTORS AT ROCKY MOUNTAIN ARSENAL
NATIONAL WILDLIFE AREA
Introduction
Prairie ecosystems typically support numerous breeding and wintering
raptors. Olendorff (1972, 1973) described the ecology and reproduction of raptor
species at the Pawnee National Grasslands in northeastern Colorado, but few studies
have been conducted on raptors in the Denver-metro region. Habitat use or selection
by diurnal raptors has been studied extensively (e.g., Janes 1985, Craighead and
Craighead 1956, Kimsey and Conley, Schmutz 1989, Per Widen 1994). Most of
these studies on raptor habitat use have focused on the reproductive period (but see
Preston 1990). Objectives of this portion of my study were to:
1. Document year-round habitat use and preference of all common diurnal
raptors at RMA.
2. Develop an ordination of coexisting raptor species that describes how
these species distribute themselves with respect to specific habitat
components.
34


In order to accomplish these objectives, I tested the null hypothesis that
diurnal raptors at RMA use habitat in direct proportion to its availability.
Methods
Habitat Classification
Habitat is defined for this portion of my study as a spatially contiguous and
primarily homogenous vegetation type that is distinctive from other such types
(Partridge 1978, Hutto 1985). A landscape composition map (Figure 3.1) was
generated from existing maps created on an Arc-info Geographic Information System
(GIS) by Morrison-Knudson Environmental Inc. (MKE). Based on structural
characteristics, vegetation map units delineated on the MKE map were combined
into eight ecologically important vegetation types. These vegetation types were:
Shortgrass steppe (including sand dropseed and needle-and-thread grass)
Shrub grassland (yucca and sand sagebrush )
Weedy forb
Crested wheatgrass
Wetland
Woodland/Riparian
Disturbed/Unvegetated
Cropland (in areas adjacent to RMA)
35


LEGEND
ON
CXBSIED WHEATCRASS
DISTURBED AREAS
WEEDY FORBS
MINOR VBGBEAHON TYPES
RIPARIAN
SEEDED AREAS
9JRUBLANX
UNCLASSHED
WATER
GRASSLANDS
WETLANDS
PH.DogTb.
t
-K-
I
SCALE 1
1
54,000
2
7FE3Z3
km
Figure 3.1 Rocky Mountain Arsenal landscape composition
o


An area was characterized as woodland if a minimum of five trees greater
than 10 meters in height were in proximity (canopies contiguous or nearly
contiguous). Elms, poplars, and other non-riparian tree species are depicted as minor
vegetation types on the map.
Cropland consisted of agricultural fields, primarily of winter wheat, to the
east of RMA. Cropland areas within 400 m of the eastern perimeter were included in
roadside counts because they are occasionally used by raptors. Seeded areas were
classified as either shortgrass or disturbed depending on seed compositions and
development stage. Open water habitats were assumed to be unavailable to raptors
because these habitats were essentially frozen all winter and unused by both winter
and summer raptors.
I determined habitat availability by using the GIS to calculate the area of each
dominant vegetation type in square kilometers (km ) for the entire study area and
each individual roadside count segment (approximately 1600 by 800 m). I then
summed the percentage of each vegetation type available within each segment to
obtain the percentage of habitat available over the entire roadside transect corridor.
Each roadside count segment was annually ground checked to verify vegetation
classification and record changing landscapes (e.g. expanding prairie dog towns,
remediation, revegetation). I determined habitat use by locating each raptor on the
37


landscape composition map (Figure 3.1) and recording dominant vegetation in a 100
m diameter circle surrounding each bird observed.
The most prominent habitat type within the road count corridor was weedy
forb which accounted for 36 % of the landscape. The next most prevalent habitat
types consisted of shortgrass (20 %) and crested wheatgrass (19.5 %). The
remaining five habitat types accounted for between 2 and 6 % of the landscape
composition. For purposes of habitat use analysis, each dominant vegetation
component was considered a distinct habitat available to diurnal raptors.
Roadside Counts
Observations of diurnal raptors collected during roadside counts were
summarized by the number of observations in specific habitat types. Habitat use-
availability analysis following the technique developed by Neu et ah (1974) and
refined by Byers and Steinhorst (1984) was conducted for all road count
observations. This procedure involves two steps. First, a chi-square goodness of fit
test of the hypothesis that raptors use habitat in proportion to availability is
performed, Second, if the null hypothesis is rejected, the preference or avoidance of
each habitat type is determined based on Bonferroni z confidence intervals
constructed around observed proportion of habitat use (Byers and Steinhorst 1984).
38


In order to use the goodness of fit approximation, the average (over all categories) of
expected observations should be greater than six (Neu et al. 1974).
Principal Component Analysis
After determining macro habitat selection of diurnal raptors at RMA based on
landscape composition, I used principal component analysis (PCA) to provide insight
into how raptors distribute themselves with respect to habitat components. I separated
road count data into summer and winter periods and analyzed 11 habitat variables
potentially selected by the primary raptors seasonally occupying RMA. Spring and fall
migration periods were not analyzed due to yearly variability. The primary summer
residents at RMA were American kestrel, red-tailed hawk and Swainsons hawk.
Primary winter residents at RMA were red-tailed hawk, rough-legged hawk,
ferruginous hawk and bald eagle. Habitat variables measured included five types of
perch substrates (tree, shrub, telephone pole, fence (8 ft. chain-link) and other human
structures), five landscape composition types (crested wheatgrass (CW), shortgrass
steppe(SG), shrubland (SB), wooded/riparian(WD) and disturbed/bare ground (UB)),
and distance to roads. Distance to road was a perpendicular distance from the raptor to
the nearest section road. The following data were recorded for each raptor observed
during road transects:
39


Number of each perch type within 50 m radius of the raptor.
Percentage of landscape composition type within a 50 m radius centered on
the raptor (measured by mapping the UTM location of each raptor
observed on the GIS vegetation map and delineating a 100 m diameter
circle around each location).
Distance to nearest section road in meters.
Because of the difficulty in distinguishing between trees as a perch substrate and
wooded/riparian as a vegetation cover type during PCA, I combined and simplified
perch variables into a single perch variable (total number of perches within a 100 m
diameter circle centered on the observed raptor). Percentages were converted using
arcsin square root transformation tables.
Principal components analysis reduces the dimensions of a complex group of
data by producing a smaller number of abstract variables. This procedure ordinates
species along axes that maximizes variance of the linear combinations of variables. I
conducted PCA of the correlation matrix of the reduced variable set using the
PRINCOMP procedure in SAS (SAS Institute, 1985). I then totaled and averaged
scores of the principal components that accounted for the greatest variance in habitat
use. I used Harvard Graphics (Bitstream Inc. 1991) to plot two-dimensional plots of
mean PCA scores plotted against each other.
40


Results
Analysis of Habitat Use and Availability From Roadside Counts
I observed six species of diurnal raptors (bald eagle, ferruginous hawk, red-
tailed hawk, rough-legged hawk, Swainsons hawk, and American kestrel)
frequently enough during roadside counts for analysis of habitat use and availability.
Other diurnal raptors, including prairie falcon, peregrine falcon, merlin, golden
eagle, northern harrier, turkey vulture, and Coopers hawk were observed on average
less than once per roadside count and were not further evaluated.
Goodness of fit comparisons of roadside count data revealed that the actual
number of all diurnal raptors observed within each habitat type differed significantly
from the expected number of observations based on the occurrence of habitats within
the study area (X2 = 615.2, df = 7, P < 0.001). The null hypothesis was therefore
rejected, implying that diurnal raptors at RMA did not use habitats in proportion to
availability. Confidence intervals revealed that raptors preferred (used more than
expected) weedy forb, wooded, and wetland areas and avoided (used less than
expected) all other habitat types (Table 3.1). To further understand raptor habitat
use, I analyzed goodness of fit comparisons by season on a species-specific basis.
Summer breeding diurnal raptors at RMA consisted of primarily red-tailed
hawk, Swainsons hawk and America kestrel. Goodness of fit comparisons of
41


Table 3.1 Habitat distribution of all diurnal raptors at Rocky Mountain Arsenal NWA as recorded by roadside counts total observations, 1991-94
Habitat Type Proportion of Study Area Number of Observations Number of Observation Expected Chi-square Contribution Proportion of Observations Pi Z1 Confidence Interval (90%) Lower Upper Limit Limit Observed Use Different From Expected2
Crested Wheatgrass 0.195 157.5 252.34 35.643 0.122 0.023 0.099 0.145 less
Weedy Forb 0.362 543.5 467.56 12.336 0.420 0.034 0.386 0.455 more
Short Grass 0.201 204 259.91 12.026 0.158 0.025 0.132 0.183 less
Shrubland 0.057 42 73.95 13.806 0.032 0.012 0.020 0.045 less
Wooded 0.023 112.5 29.80 229.514 0.087 0.020 0.067 0.107 more
Cropland 0.068 34 88.50 33.565 0.026 0.011 0.015 0.037 less
Urban/Disturbed 0.059 45 75.88 12.565 0.035 0.013 0.022 0.048 less
Wetland 0.035 154.5 45.07 265.739 0.119 0.023 0.097 0.142 more
Total 1.000 1293 1293 615.194 1.000
Z(l-a/2k) 2.5
1 = Z(l-a/2K) x (Pi(l-Pi)/n) I/2, K = number of habitat categories
2 = Observed use different from expected when 90 % confidence intervals around observed use falls below (less than expected)
or above (more than expected) proportion of habitat within the study area


roadside count data for these three species revealed that red-tailed and Swainsons
hawks essentially used habitats in proportion to availability, although Swainsons
hawks used shrubland less than expected (Tables 3.2 and 3.3). However, the reader
should consider the results from red-tailed hawks with caution because the small
number of expected observations (44) is less than an average of six over all
categories and violates assumptions of goodness of fit tests. Summer kestrels used
weedy forb habitats more than expected and shortgrass, crested wheatgrass and
cropland habitats less than expected (Table 3.4).
Wintering raptors at RMA consisted primarily of bald eagles, ferruginous,
rough-legged, and red-tailed hawks. Golden eagles, northern harriers and American
kestrels also occurred in winter but not frequently enough for analysis. Goodness of
fit comparisons of roadside count data for the four primary wintering raptors revealed
that none of these species used habitats in proportion to availability. The number of
bald eagles observed within each habitat type differed significantly from the
expected number of observations based on the occurrence of habitats within the
study area (.X2 = 373.04, df= 7, P < 0.001). Bald eagles used wooded and wetland
areas more than expected, while using crested wheatgrass, shortgrass, shrubland, and
urban/disturbed areas less than expected (Table 3.5). Bald eagles were not observed
in croplands. Wintering ferruginous hawks observed within each habitat type
differed significantly from the expected number of observations (A2 = 53.7, df = 7, P
43


Table 3.2 Habitat distribution of red-tailed hawks at Rocky Mountain Arsenal NWA as recorded by roadside counts summer, 1991 93
Habitat Type Proportion of Study Area Number of Observation Number of Observations Expected Chi-square Contribution Proportion of Observations Pi Z1 Confidence Interval (90%) Observed Use Lower Upper Different From Limit Limit Expected2
Crested Wheatgrass 0.195 7 8.587 0.293 0.159 0.138 0.021 0.297
Weedy Forb 0.362 22 15.911 2.331 0.500 0.188 0.312 0.688
Short Grass 0.201 6 8.845 0.915 0.136 0.129 0.007 0.266
Shrubland 0.057 5.5 2.517 3.537 0.125 0.125 0.000 0.250
Wooded 0.023 0.5 1.014 0.261 0.011 0.040 -0.029 0.051
Cropland 0.068 0 3.012 3.012 0.000 0.000 0.000 0.000
Urban/Disturbed 0.059 0 2.582 2.582 0.000 0.000 0.000 0.000
Wetland 0.035 3 1.534 1.402 0.068 0.095 -0.027 0.163
Total 1.000 44 44 14.332 1.000
Z(l-a/2k) 2.5
1 = Z(l-a/2K) x (Pi(l-Pi)/n) l/2, K = number of habitat categories
2 = Observed use different from expected when 90 % confidence intervals around observed use falls below (less than expected)
or above (more than expected) proportion of habitat within the study area, blank spaces indicates habitat use
was in proportion to availability


Table 3.3 Habitat distribution of Swainsons hawks at Rocky Mountain Arsenal NWA as recorded by roadside counts summer, 1991 93
Proportion Number of Proportion of Confidence Interval (90%) Observed Use
Habitat Type of Study Area Number of Observation Observations Expected Chi-square Contribution Observations Pi z1 Lower Limit Upper Limit Different From Expected2
Crested Wheatgrass 0.195 9 14.246 1.932 0.123 0.096 0.027 0.219
Weedy Forb 0.362 35.5 26.397 3.139 0.486 0.146 0.340 0.633
Short Grass 0.201 14.5 14.674 0.002 0.199 0.117 0.082 0.315
Shrubland 0.057 1 4.175 2.415 0.014 0.034 -0.020 0.048 less
Wooded 0.023 7 1.682 16.807 0.096 0.086 0.010 0.182
Cropland 0.068 4 4.997 0.199 0.055 0.067 -0.012 0.121
Urban/Disturbed 0.059 2 4.284 1.218 0.027 0.048 -0.020 0.075
Wetland 0.035 0 2.544 2.544 0.000 0.000 0.000 0.000
Total 1.000 73 73 28.256 1.000
Z(l-a/2k) 2.5
1 = Z(l-a/2K) x (Pi(l-Pi)/n) 1/2, K = number of habitat categories
2 = Observed use different from expected when 90 % confidence intervals around observed use falls below (less than expected)
or above (more than expected) proportion of habitat within the study area


Table 3.4 Habitat distribution of American kestrels at Rocky Mountain Arsenal NWA as recorded by roadside counts summer, 1991 93
Proportion Number of Proportion of Confidence Interval (90%) Observed Use
Habitat Type of Study Area Number of Observation Observations Expected Chi-square Contribution Observations Pi z1 Lower Limit Upper Limit Different From Expected2
Crested Wheatgrass 0.195 11.5 20.687 4.080 0.108 0.076 0.033 0.184 less
Weedy Forb 0.362 64 38.330 17.191 0.604 0.119 0.485 0.723 more
Short Grass 0.201 8.5 21.307 7.698 0.080 0.066 0.014 0.146 less
Shrubland 0.057 3 6.063 1.547 0.028 0.040 -0.012 0.069
Wooded 0.023 6 2.443 5.179 0.057 0.056 0.000 0.113
Cropland 0.068 1.5 7.255 4.566 0.014 0.029 -0.015 0.043 less
Urban/Disturbed 0.059 2.5 6.220 2.225 0.024 0.037 -0.013 0.060
Wetland 0.035 9 3.695 7.619 0.085 0.068 0.017 0.153
Total 1.000 106 106 50.105 1.000
Z)l-a/2k)__________________________________________________ 2.5
1 = Z(l-a/2K) x (Pi(l-Pi)/n) l/2, K = number of habitat categories
2 = Observed use different from expected when 90 % confidence intervals around observed use falls below (less than expected)
or above (more than expected) proportion of habitat within the study area


Table 3.5 Habitat distribution of bald eagles at Rocky Mountain Arsenal NWA as recorded by roadside counts winter, 1991 94
Habitat Type Proportion of Study Area Number of Observations Number of Observations Expected Chi-square Contribution Proportion of Observations Pi z1 Confidence Interval (90%) Lower Upper Limit Limit Observed Use Different From Expected2
Crested Wheatgrass 0.195 1 14.832 12.899 0.013 0.033 -0.020 0.046 less
Weedy Forb 0.362 19.5 27.482 2.318 0.257 0.125 0.131 0.382
Short Grass 0.201 6.5 15.277 5.043 0.086 0.080 0.005 0.166 less
Shrubland 0.057 0.5 4.347 3.404 0.007 0.023 -0.017 0.030 less
Wooded 0.023 24 1.752 282.603 0.316 0.133 0.182 0.449 more
Cropland 0.068 0 5.202 5.202 0.000 0.000 0.000 0.000
Urban/Disturbed 0.059 0.5 4.460 3.516 0.007 0.023 -0.017 0.030 less
Wetland 0.035 24 2.649 172.098 0.316 0.133 0.182 0.449 more
Total 1.000 76 76 487.083 1.000
Z(l-a/2k) 2.5
1 = Z(l-a/2K) x (Pi(l-Pi)/n) 1/2, K = number of habitat categories
2 = Observed use different from expected when 90 % confidence intervals around observed use falls below (less than expected)
or above (more than expected) proportion of habitat within the study area


< 0.001). The null hypothesis was therefore rejected, implying that ferruginous
hawks wintering at RMA did not use habitats in proportion to availability. Applying
confidence intervals revealed that ferruginous hawks preferred weedy forb areas and
avoided crested wheatgrass, cropland, and wetland areas (Table 3.6). The number of
wintering rough-legged hawks observed also differed significantly from the expected
number of observations (X2 = 66.6, df = 7, P < 0.001). Rough-legged hawks
preferred wetland areas, avoided cropland and shortgrass steppe, and used all other
habitats in proportion to availability (Table 3.7). The number of wintering red-
tailed hawks observed also differed significantly from the expected number of
observations (X2 = 515.9, df = 7,P< 0.001). Red-tailed hawks preferred wooded
and wetland areas and avoided crested wheatgrass, weedy forb, shortgrass, shrubland
and cropland areas (Table 3.8).
Principal Components Analysis of Roadside Raptor Counts
Summer. The first two principal components accounted for 50.1 % of the
total variance observed in summer roadside counts. The first principal component
(PC 1) accounted for 25.76 % of the variance (Table 3.9). The highest positive
correlation of the original variables (loading) is percentage of crested wheatgrass,
and the highest negative correlation is percentage of shortgrass steppe. This
represents a gradient of vegetation diversity from more diverse shortgrass steppe to
48


Table 3.6 Habitat distribution of ferruginous hawks at Rocky Mountain Arsenal NWA as recorded by roadside counts winter, 1991 94
Habitat Type Proportion of Study Area Number of Observations Number of Observations Expected Chi-square Contribution Proportion of Observations Pi Z1 Confidence Interval (90%) Lower Upper Limit Limit Observed Use Different From Expected2
Crested Wheatgrass 0.195 16 36.494 11.509 0.086 0.051 0.034 0.137 less
Weedy Forb 0.362 109.5 67.620 25.938 0.586 0.090 0.496 0.676 more
Short Grass 0.201 30 37.589 1.532 0.160 0.067 0.093 0.228
Shrubland 0.057 7.5 10.695 0.955 0.040 0.036 0.004 0.076
Wooded 0.023 5.5 4.310 0.329 0.029 0.031 -0.001 0.060
Cropland 0.068 2 12.800 9.112 0.011 0.019 -0.008 0.030 less
Urban/Disturbed 0.059 14.5 10.974 1.133 0.078 0.049 0.029 0.126
Wetland 0.035 2 6.518 3.131 0.011 0.019 -0.008 0.030 less
Total 1.000 187 187.000 53.639 1.000
Z(l-a/2k) 2.500
1 = Z(l-a/2K) x (Pi(l-Pi)/n) 1/2, K = number of habitat categories
2 = Observed use different from expected when 90 % confidence intervals around observed use falls below (less than expected)
or above (more than expected) proportion of habitat within the study area


Table 3.7 Habitat distribution of rough-legged hawks at Rocky Mountain Arsenal NWA as recorded by roadside counts winter, 1991 94
Habitat Type Proportion of Study Area Number of Observations Number of Observations Expected Chi-square Contribution Proportion of Observations Pi z' Confidence Interval (90%) Lower Upper Limit Limit Observed Use Different From Expected1 2
Crested Wheatgrass 0.195 7.5 13.856 2.916 0.106 0.091 0.014 0.197
Weedy Forb 0.362 31.5 25.674 1.322 0.444 0.147 0.296 0.591
Short Grass 0.201 7.5 14.272 3.213 0.106 0.091 0.014 0.197 less
Shrubland 0.057 4 4.061 0.001 0.056 0.068 -0.012 0.125
Wooded 0.023 4.5 1.636 5.012 0.063 0.072 -0.009 0.136
Cropland 0.068 0.5 4.860 3.911 0.007 0.025 -0.018 0.032 less
Urban/Disturbed 0.059 2 4.166 1.127 0.028 0.049 -0.021 0.077
Wetland 0.035 13.5 2.475 49.122 0.190 0.116 0.074 0.307 more
Total 1 71 71.000 66.623 1.000
Z( 1 -a/2k) 2.5
1 = Z(l-a/2K) x (Pi(l-Pi)/n) 1/2, K = number of habitat categories
2 = Observed use different from expected when 90 % confidence intervals around observed use falls below (less than expected)
or above (more than expected) proportion of habitat within the study area


Table 3.8. Habitat distribution of red-tailed hawks at Rocky Mountain Arsenal NWA as recorded by roadside counts winter, 1991 94
Habitat Type Proportion of Study Area Number of Observations Number of Observations Expected Chi-square Contribution Proportion of Observations Pi Z1 Confidence Interval (90%) Lower Upper Limit Limit Observed Use Different From Expected2
Crested Wheatgrass 0.195 11 22.833 6.133 0.094 0.067 0.027 0.161 less
Weedy Forb 0.362 26 42.308 6.286 0.222 0.096 0.126 0.318 less
Short Grass 0.201 12.5 23.518 5.162 0.107 0.071 0.035 0.178 less
Shrubland 0.057 1.5 6.692 4.028 0.013 0.026 -0.013 0.039 less
Wooded 0.023 33 2.696 340.558 0.282 0.104 0.178 0.386 more
Cropland 0.068 1.5 8.008 5.289 0.013 0.026 -0.013 0.039 less
Urban/Disturbed 0.059 3 6.866 2.177 0.026 0.037 -0.011 0.062
Wetland 0.035 28.5 4.078 146.261 0.244 0.099 0.144 0.343 more
Total 1.000 117 117 515.894 1.000
Z(l-a/2k) 2.5
1 = Z(l-a/2K) x (Pi(l-Pi)/n) 1/2, K = number of habitat categories
2 = Observed use different from expected when 90 % confidence intervals around observed use falls below (less than expected)
or above (more than expected) proportion of habitat within the study area


Table 3.9 Summary statistics of a principal components analysis of diurnal raptor
_______________summer habitat components on Rocky Mountain Arsenal, NWA____________________________
Principal components
Statistic 1 2 3 4
Eigenvalue 1.8 1.7 1.12 1.05
% of variance 25.76 24.34 16.05 15.01
Cumulative % 25.76 50.1 66.15 81.16
Component Eigenvectors
Perch 0.265 0.599 -0.107 -0.154
CW 0.485 -0.457 -0.293 -0.334
SG -0.694 0.161 -0.162 0.082
SB 0.084 -0.049 0.930 -0.019
WD 0.268 0.630 -0.006 -0.089
UB 0.205 -0.052 -0.104 0.844
RD 0.303 0.067 0.009 0.371
Perch = Number of perches (trees, poles, structures) within 100m diameter circle of observed raptor
CW = crested wheatgrass, SG = shortgrass steppe, SB = shrubland
WD = Wooded/Riparian, UB = Urban /unvegetated.
RD = Distance to nearest road
52


monotypic crested wheatgrass habitat. Species found in more monotypic and
introduced habitats would be expected to have high PC 1 values. Species found in
more diverse habitats would be expected to have low PC 1 values.
The second principal component (PC 2) accounted for an additional 24.34 %
of the variance observed in summer roadside counts (Table 3.9). The highest
positive loadings are the number of perches and the percentage of wooded/riparian
habitat. The highest negative loading is the percentage of crested wheatgrass habitat.
This component represents increasing woodland and number of perches. Species
found in wooded areas with numerous perch sites would be expected to have high PC
2 values. Species found in habitats with little woodland/riparian vegetation and few
perches would be expected to have low PC 2 values.
Habitat relationships of diurnal raptors on RMA can be reconstructed by
using the principal components as coordinates in a two-dimensional plot (James
1971). Figure 3.2 presents the relationship of summer diurnal raptors along the axis
of the first two principal components. The horizontal axis, representing PC 1, has
separated the species along a vegetation diversity gradient. Swainsons hawk
occurred in more monotypic crested wheatgrass habitat than red-tailed hawk and
especially American kestrel, which occurred in more diverse habitats (Figure 3.2).
The vertical axis, representing PC 2, separated the species along a gradient of
increasing perches and wooded/riparian habitat. Red-tailed hawks occurred in
53


'wO
Wooded /Riparian 0*3
Perches
: 0.25
CM
O
Q_
D
c

T3
O
o
*
oo

Q.
00
c
CO
0.2 I
j
0.15
0.1
0.05
-0.05
AK
A
RTH
Â¥
SH

-0.8 -0.6 -0.4 -0.2 0
0.2 0.4 0.6 0.8
1
Diverse
vegetation
PC 1
Vegetation diversity
Monotypic non-native
> vegetation
Figure 3.2 Two-dimensional ordination of three species of summer diurnal raptors at Rocky Mountain Arsenal NWA
according to the first two principal components (PC I and PC 2)
Locations of raptors along plot axes represents mean principal component scores.
AK = American kestrel RTH = Red-tailed hawk, SH = Swainson's hawk


habitats with more perches and woodland than either kestrels or Swainsons hawks.
American kestrels occurred in areas with the fewest number of perches and little
woodland habitat (Figure 3.2).
Winter. The first two principal components accounted for 49.0 % of the total
variance observed during winter roadside counts. The first principal component
accounted for 29.1 % of the variance (Table 3.10). The highest positive loadings are
the percentage of wooded/riparian habitat and number of perches. The only negative
loading is the percentage of shortgrass steppe habitat. This represents a gradient of
increasing wooded/riparian habitat and perch availability, and decreasing grassland.
Species occurring in more wooded habitats with numerous perches would be
expected to have high mean PC 1 values. Species occurring in grassland habitats
with few perches would be expected to have low mean PC 1 values.
The second principal component accounted for an additional 19.9 % of the
variance observed during winter roadside counts. The highest positive loadings are
percentage of shortgrass steppe and wooded/riparian habitats. The most negative
loadings are percentage of crested wheatgrass and urban/unvegetated habitats (Table
3.10). This represents a gradient of vegetation cover and diversity from disturbed
bare ground habitats to more native and diverse habitats. Species occurring in
diverse native habitats would be expected to have high PC 2 values. Species
55


Table 3.10 Summary statistics of a principal components analysis of diurnal raptor
____________overwintering habitat components on Rocky Mountain Arsenal, NWA
Principal components
Statistic 1 2 3 4
Eigenvalue 2.04 1.39 1.08 1.04
% of variance 29.13 19.89 15.39 14.82
Cumulative % 29.13 49.02 64.41 79.23
Component Eigenvectors
Perch 0.516 0.261 -0.167 -0.094
CW 0.101 -0.749 -0.300 0.087
SG -0.595 0.393 -0.143 -0.019
SB 0.261 0.016 -0.084 0.702
WD 0.548 0.317 0.070 -0.282
UB 0.015 -0.263 0.848 -0.162
RD 0.004 0.215 0.361 0.620
Perch = Number of perches (trees, poles, structures) within 100m diameter circle of observed raptor
CW = crested wheatgrass, SG = shortgrass steppe, SB = shrubland
WD = Wooded/Riparian, UB = Urban /unvegetated.
RD = Distance to nearest road
56


occurring in monotypic crested wheatgrass and disturbed, unvegetated habitats
would be expected to have low PC 2 values.
Figure 3.3 presents the relationship of winter diurnal raptors along the axis of
the first two principal components. The horizontal axis, representing PC 1, has
separated the species along a gradient of increasing woodland and number of
perches. The bald eagle and red-tailed hawk produced high mean value PC 1 scores
indicating that these species occurred in wooded habitats with a high degree of perch
availability. In contrast, the ferruginous hawk produced the lowest mean PC 1 score
indicating that this species occurred in grassland habitats with a low degree of perch
availability. Rough-legged hawks occurred in grasslands with relatively more
woodlands and perches than ferruginous hawks, but with notably fewer woodlands or
perches than either red-tailed hawks or bald eagles.
The vertical axis, representing PC 2, separated wintering raptors at RMA
along a vegetation cover and diversity gradient (Figure 3.3). The bald eagle
produced the highest mean PC 2 score indicating that this species occurred in much
more diverse habitat than ferruginous, red-tailed, and rough-legged hawks. The
ferruginous hawk produced the lowest mean PC 2 score indicating that this species
occurred in habitats with relatively lower vegetation cover and diversity (Figure 3.3).
57


Diverse
vegetation
CM
O
0-
4<
00
o
>
TD
C
03
OJ
>
O
o
c
o
o
&0
2
Monotypic
disturbed
vegetation
Grasslands
Increasing woodland
Wooded/Riparian
Figure 3.3 Two-dimensional ordination of four species of winter diurnal raptors at Rocky Mountain Arsenal NWA
according to the first two principal components (PC 1 and PC 2)
Locations of raptors along plot axes represents mean principal component scores.
AK = American kestrel, RTH = Red-tailed hawk, SH = Swainson's hawk


Discussion
The winter and summer distribution of raptors at RMA are undoubtedly
related to habitat selection. I found that breeding diurnal raptors at RMA segregated
in relation to habitat. American kestrel occurred in grasslands, red-tailed hawk in
more wooded areas with numerous perches and Swainsons hawk occupied habitats
in-between. Schmutz et al. (1980) in a study of sympatric buteos breeding in
grassland habitats in Canada found a similar separation among species, with red-
tailed hawks associated with wooded parkland; ferruginous hawks with open prairie;
and Swainson's hawk occupying the ecotone between wooded parkland and open
prairie. Janes (1985) in a similar study found red-tailed hawks in areas associated
with relatively high perch density, ferruginous hawks in areas of relatively flat,
unbroken habitat without perches, and Swainson's hawks in relatively flat areas
containing a few, widespread perches. Smith and Murphy (1973) found similar
habitat relationships in Utah, and also found that topographic relief and specific plant
associations were important to hawk distribution.
Summer and winter raptors at RMA select habitats disproportionate to
availability. Summer American kestrel preferred weedy areas while avoiding
shortgrass and cropland areas. This finding is consistent with distribution patterns
described in Chapter 2. Summer red-tailed hawks and Swainsons hawks generally
59


used habitats in proportion to availability but Swainsons hawks avoided shrubland
areas. Although shrubland areas support some of the highest densities of small
mammals at RMA (Boone and Preston 1995) the structure and height of the
vegetation may make prey unavailable to buteos (Preston 1990, Wakely 1978, Baker
and Brooks 1981, Bechard 1982).
Integrating raptor distribution results with habitat use and availability results
provides a clearer picture of winter raptor ecology at RMA. Wintering ferruginous
hawks prefer weedy forb habitats, primarily in the northeastern and central regions of
RMA. Red-tailed hawks prefer wooded and wetland habitats primarily in the
southwestern and southeastern regions of RMA which contains the highest
percentage of these habitats. Bald eagles also prefer wooded and wetland habitats,
primarily associated with the lakes and First Creek in the southeastern and
northeastern regions of RMA.
Measurements from roadside counts are organized by individual species
observed with little attention paid to other nearby species. This approach assumes
that predictable relationships exist between the occurrence of a bird and its
characteristic vegetation and/or habitat requirements, or niche-gestalt (James 1971).
Inherent to the term gestalt are the concepts that each species has a set of proximate
factors that it responds to, and that it has a predetermined set of specific search
images (Tinbergen 1951). The habitat gestalt for summer species of diurnal raptors
60


at RMA determined by combining PCA and habitat use/availability analysis provides
a clear picture of breeding raptor habitat gestalts. American kestrels occurred in
diverse vegetation with few perches, particularly weedy forb areas that typically
support abundant insect populations. Red-tailed hawks most frequently occurred in
disturbed central regions of RMA associated with non-native vegetation and
adequate trees for perching. Swainsons hawks also occurred in monotypic, non-
native vegetation, but with fewer perches and woodland areas than red-tailed hawks.
The gestalt for wintering raptors at RMA was fairly evident for the four
common species. Both PCA and habitat use and availability analysis indicated a
gestalt of grassland habitats generally lacking perches for wintering ferruginous
hawks at RMA. These results confirm the findings of Janes (1985) and Bechard and
Schmutz (1995). The gestalt for both bald eagles and red-tailed hawks wintering at
RMA was wooded habitats with numerous perches. However, bald eagles occurred
in areas with greater diversity and vegetation cover. Rough-legged hawks appeared
to use most habitat types, although they tended to prefer wetland areas and avoid
shortgrass and cropland areas.
Summer American kestrels and winter ferruginous hawks both preferred weedy
forb habitat. However, kestrels used habitats with high vegetation cover and diversity;
whereas, wintering ferruginous hawks used areas with low vegetation cover and
diversity. This apparent contradiction can be explained by the species selecting for
61


specific habitat characteristics within the more general weedy forb type. American
kestrels selected weedy forb areas that contained diverse mosaics of weedy forbs and
grasslands. These areas support high concentrations of prey (e.g., grasshoppers and
grassland birds) in summer that likely attract American kestrels. However, wintering
ferruginous hawks selected low-growing weedy areas with little diversity, specifically
prairie dog towns.
Although PCA and habitat use/availability analysis provided a habitat gestalt
for summer and winter raptors at RMA, other community factors, such as
competition and prey availability, may have been missed in the analyses that could
also influence habitat use (see James 1971). Vegetation structure has also been
associated with habitat selection and distribution in raptors and was not specifically
investigated during this study. However, Janes (1985) and Preston (1990) found that
vegetation structure alone did not exert a strong influence on habitat occupied by
hawks. Prey related habitat selection has also been suggested for red-tailed and
rough-legged hawks (Schnell 1968) and ferruginous hawks (Howard and Wolfe
1976, Lardy 1980). Prey abundance and availability components could have
strengthened my analysis, particularly for summer data. Several studies (Preston
1990, Baker and Brooks 1981, Schmutz and Hungle 1989) have found that raptor
distribution and territory size is influenced by prey. This component may be critical
for breeding raptors that need abundant and available prey to feed young.
62


CHAPTER4
WINTER ECOLOGY OF FERRUGINOUS HAWKS AT ROCKY MOUNTAIN
ARSENAL NATIONAL WILDLIFE AREA
Introduction
The ferruginous hawk is currently classified as a candidate species for the
federal Threatened and Endangered Species List under the Endangered Species Act of
1973 (16 U.S.C.). The primary reason for considering the listing is a continued decline
in populations resulting from habitat destruction and reduced prey populations (caused
by agricultural practices) (USFWS 1991). Olendorff (1993) confirmed population
declines only in Utah and Nevada and attributed these declines primarily to cultivation.
Grazing, small mammal control programs, mining and fire also contribute to
ferruginous hawk declines (Olendorff 1993). Recent emphasis on conservation of
wintering grounds and migration corridors for migrant species (Terborgh 1992) has
underscored the importance of identifying essential habitat components in all
ecosystems in which a species occurs. As a migratory raptor, the overwintering
condition of ferruginous hawks is important to the overall reproductive rate of the
species. The goal of this portion of my study was to identify and describe several
63


ecological aspects of overwintering ferruginous hawks at RMA, including detailed
habitat use, home range, and characteristics of communal roosts.
Although numerous investigators have described the habitat use of breeding
ferruginous hawks and other sympatric prairie species of raptors (Schmutz et al. 1980,
Cotrell 1981, Bechard et al. 1990, Restani 1991), little information is available on the
habitat use and home range of wintering ferruginous hawks. Communal roost use by
wintering ferruginous hawks has also rarely been reported in the literature (but see
Steenhof 1984). Ferruginous hawk communal roost sites first identified at RMA by
M. Lockhart (USFWS) in 1988, are further described in this portion of my study.
Methods
Habitat Use
Habitat use by wintering ferruginous hawks at RMA was analyzed using data
collected by two different methods; roadside counts and radio-tracking. I analyzed
habitat use and availability from both radio-tracking and roadside count data
according to the techniques of Neu et al. (1974) and Byers and Steinhorst (1984) as
described in Chapter 3.
Radio-Tracking Twenty-eight ferruginous hawks were trapped at RMA
using the modified Lockhart method (Bloom 1987) during three consecutive winters
from 1991 through 1994. I attached tail-mounted radio transmitters to the central
64


rectrix of captured hawks and secured the transmitter and antenna with thread and
epoxy (Kenward 1987). Sufficient data to analyze habitat use/availability were
collected for 17 radio-tagged hawks. I summarized the total amount of time
ferruginous hawks were observed in each respective habitat type. Analyses were
restricted to observations of radio-tagged hawks within an area that consisted of
RMA and a buffer area that included adjacent portions of Stapleton International
Airport (Sections 9 and 10), and a 400m corridor along the eastern boundary. This
buffer zone corresponded to areas covered by the roadside count transect. The habitat
available to hawks was assumed to consist of all habitats within RMA and the buffer
area.
For purposes of habitat use analysis, each dominant vegetation component
was considered a distinct habitat available to ferruginous hawks. Prairie dog
distribution was calculated based on maps and estimates compiled by the USFWS on
a yearly basis. The distribution of prairie dogs within each habitat component
consisted of the average extent of prairie dogs over the three years of study.
Roadside Counts. I observed ferruginous hawks during roadside counts as
described in the previous chapter for diurnal raptors. Observations of flying
ferruginous hawks were included in relative abundance estimates described in
Chapter 2, but were excluded from habitat use and availability analysis.
65


Home Range of Wintering Ferruginous Hawks
Radio-tagged ferruginous hawks were tracked during fall and winter 1991-92,
1992-93, and 1993-94. During the first winter of study, radio-tagged birds were
tracked for two 4-hour periods per week (systematic relocations) (Andersen and
Rongstad 1989). During the final two winters of study, radio-tagged birds were
tracked for a 12-hour (dawn to dusk) period every 5-7 days.
In order to minimize bias caused by capture stress, I began tracking radio-tagged
hawks two to five days after capture (White and Garrot 1990). I then tracked hawks
systematically until the individual hawks either shed the transmitter or left the area
and could not be relocated. Hawk relocations were marked on a 1:24,000
topographic maps and recorded by Universal Trans-Mercator (UTM) to the nearest
10 meters. Additionally, as many radio-tagged birds as possible were visually
relocated once per day, two to five days per week (random relocations) to help
determine spatial distribution patterns of ferruginous hawks at RMA.
Home range size for each radio-tagged ferruginous hawk was estimated using
the minimum convex polygon approach (Mohr 1947, Southwood 1966). Areas of
high diurnal ferruginous hawk use were identified using the 50% harmonic mean
activity area (Dixon and Chapman 1980). Home range analyses were conducted on
66


McPaal (Stuwe 1992) home range computer software. All non-identical, consecutive
relocations were considered data points for home range analysis.
Roost Site Characteristics
Communal roosts were defined as a single tree or a group of trees containing
more than four ferruginous hawks for a period greater than a week. Beginning in
November, 1992, observations of known ferruginous hawk communal roost sites
were conducted during dusk at least once every week to determine hawk use and
density at communal roosts. I also conducted dusk surveys for unknown roost sites
once every month. Communal roost sites identified by radio-tracking and dusk
surveys were mapped, photographed, and studied according to methods adapted from
Keister and Anthony (1983). Roost trees were measured after wintering hawks had
left communal roosts for the season and before summer breeding raptors began
nesting. Data collected from communal roosts included:
1) tree species and number of trees in the roost group,
2) diameter at breast height (DBH),
3) height classification (0-15, 16-23m, 24-30m),
4) structure classification,
5) number of roost trees within a communal roost,
6) protection from prevailing wind,
7) proximity to roads and other man-made structures, and
67


8) distance to prairie dog colonies.
Roost site fidelity was determined from radio-tracking of individual
ferruginous hawks. Fidelity was calculated based on a simple percentage of nights
observed in individual roost sites.
Results
Analysis Of Habitat Use From Radio-Tracking
Goodness of fit comparisons of radio-tracking data revealed that locations of
ferruginous hawks at RMA differed significantly from the occurrence of habitat
types within the study area (A5 = 132.7, df = 7, P < 0.001) (Table 4.1). Therefore,
the null hypothesis that ferruginous hawks used habitats in proportion to availability
was rejected. Confidence intervals indicated that radio-tagged ferruginous hawks
used shortgrass, wooded and urban/disturbed areas more than expected and used all
other habitats less than expected. Habitats were then combined into structurally
similar habitat components to determine if hawks showed similar habitat selection
based primarily on vegetation structure versus composition. The vegetation classes
combined into single habitat components included the following:
Wooded Wetland combined into a single wooded/riparian component
Crested wheatgrass Cropland combined into crested wheatgrass
component (these two habitats are structurally similar in winter and
68


Table 4.1 Habitat distribution of overwintering ferruginous hawks (n=17) at Rocky Mountain Arsenal NWA
as recorded by radio-tracking total observations, 1991 94
Habitat Type Proportion of Study Area Proportion Minutes of Total Observed Observations (Pi) Minutes Chi-square Expected Contribution Pi z1 Confidence Interval (90%) Lower Upper Limit Limit Observed Use Different From Expected2
Crested Wheatgrass 0.191 1114.00 0.031 6833.19 13.375 0.031 0.002 0.029 0.033 less
Weedy Forb 0.336 10466.50 0.292 12035.45 0.571 0.292 0.006 0.286 0.298 less
Short Grass 0.240 13549.00 0.379 8601.84 7.950 0.379 0.006 0.372 0.385 more
Shrubland 0.063 323.50 0.009 2255.25 4.623 0.009 0.001 0.008 0.010 less
Wooded 0.024 6434.50 0.180 849.61 102.577 0.180 0.005 0.175 0.185 more
Cropland 0.028 360.50 0.010 990.70 1.120 0.010 0.001 0.009 0.011 less
Urban/Disturbed 0.092 3520.00 0.098 3294.77 0.043 0.098 0.004 0.094 0.102 more
Wetland 0.026 22.00 0.001 929.19 2.475 0.001 0.000 0.000 0.001 less
Total 1.000 35790.00 1.000 35790.00 132.734 1.000
Z(l-a/2K) 2.500
1 = Z(l-a/2K) x (Pi(l-Pi)/n) 1/2, K = number of habitat categories
2 = Observed use different from expected when 90 % confidence intervals around observed use falls below (less than expected)
or above (more than expected) proportion of habitat within the study area


consist of monotypic cultivars, rigid stubble, and a high percentage of
bare ground)
Disturbed (unchanged)
Weedy forb Shortgrass combined into shortgrass component
Shrubland (unchanged)
These five habitat components were further evaluated based on the presence
and absence of prairie dogs. Habitat use-availability analysis of these combined
habitats with and without prairie dogs showed that ferruginous hawk did not use
habitats in proportion to availability (A^ = 373.0, df = 9, P < 0.001) (Table 4.2).
All habitat components, with the exception of shrubland, that had prairie dogs were
used more than expected by ferruginous hawks. Conversely the only habitat
component without prairie dogs that was used more than expected by ferruginous
hawks was the wooded/riparian type. All other habitats without prairie dogs were
used less than expected.
Analysis of Habitat Use From Roadside Counts
Goodness of fit comparisons of roadside count data showed that ferruginous
hawks preferred (used more than expected) weedy forb areas and avoided (used less
than expected) crested wheatgrass, cropland, and wetland areas (Table 3.6). Contrary
to results from radio-tracking, the roadside counts indicated ferruginous hawks
70


Table 4.2 Habitat distribution of overwintering ferruginous hawks (n=17) at Rocky Mountain Arsenal NWA
as recorded by radio-tracking total observations evaluated with the presence of prairie dogs, 1991 94
Habitat Type Proportion of Study Area Proportion Minutes of Total Observed Observations (Pi) Minutes Chi-square Expected Contribution Pi z1 Confidence Interval (90%) Lower Upper Limit Limit Observed Use Different From Expected2
Crested Wheatgrass 0.213 1085.50 0.030 7620.87 15.658 0.030 0.003 0.028 0.033 less
C. Wheatgrass w/ PD 0.003 392.00 0.011 114.40 1.882 0.011 0.002 0.009 0.013 more
Short Grass 0.485 7602.50 0.212 17343.16 15.284 0.212 0.006 0.206 0.219 less
Short Grass w/ PD 0.096 16413.00 0.459 3422.23 137.772 0.459 0.008 0.451 0.466 more
Shrubland 0.058 252.00 0.007 2086.54 4.506 0.007 0.001 0.006 0.008 less
Shrubland w/ PD 0.004 71.50 0.002 135.83 0.085 0.002 0.001 0.001 0.003 less
Wooded/Riparian 0.049 5844.50 0.163 1760.65 26.465 0.163 0.006 0.157 0.169 more
Wooded/Rip. w/ PD 0.000 612.00 0.017 10.71 94.313 0.017 0.002 0.015 0.019 more
Disturbed 0.088 1393.50 0.039 3151.38 2.740 0.039 0.003 0.036 0.042 less
Disturbed w/ PD 0.004 2126.50 0.059 147.24 74.334 0.059 0.004 0.056 0.063 more
Total 1.000 35793.00 1.000 35793.00 373.039 1.000
Z(l-a/2K) 2.576
1 = Z(l-a/2K) x (Pi(l-Pi)/n) 1/2, K = number of habitat categories
2 = Observed use different from expected when 90 % confidence intervals around observed use falls below (less than expected)
or above (more than expected) proportion of habitat within the study area


preferred weedy forb areas, although the range of radio-tracking observations was
only slightly less than the availability of weedy forb habitats. Habitats were then
combined into structurally similar habitat components to determine if hawks showed
similar habitat selection based primarily on vegetation structure versus composition.
Vegetation classes were combined into the five reduced habitat components
described above. These habitats were also evaluated based on the presence and
absence of prairie dogs.
Ferruginous hawk use of the combined habitats with and without prairie dogs
also did not occur in proportion to availability (Table 4.3). No habitat with prairie
dogs present was used less than expected. Three habitats without prairie dogs
present (crested wheatgrass, shortgrass, and shrubland) were avoided by ferruginous
hawks. Only one habitat, shortgrass with prairie dogs was preferred. Observations
of ferruginous hawks in all other habitats were in proportion to habitat availability.
These results are generally consistent with analysis of radio-tracking data. The major
difference being that fewer habitats were identified as preferred or avoided using
roadside count data.
I found that the occurrence of prairie dogs was a major environmental factor
influencing habitat selection in ferruginous hawks at RMA. Overall, shortgrass
habitats were preferred by ferruginous hawks. However, analysis of habitat use with
72


Table 4.3 Habitat distribution of overwintering ferruginous hawks on Rocky Mountain Arsenal NWA as recorded by roadside counts -
observations evaluated with the presence of prairie dogs, 1991 94
Habitat Type Proportion of Study Area Number of Observations Number of Observation Expected Chi-square Contribution Pi Confidence Interval (90%) Lower Upper Z1 Limit Limit Observed use different from Expected2
Crested Wheatgrass 0.260 14.50 48.7 24.009 0.078 0.050 0.027 0.128 less
C. Wheatgrass w/ PD 0.003 3.50 0.6 14.081 0.019 0.026 -0.007 0.044
Short Grass 0.435 56.00 81.3 7.881 0.299 0.086 0.213 0.386 less
Short Grass w/ PD 0.119 83.50 22.2 168.780 0.447 0.094 0.353 0.540 more
Shrubland 0.055 3.50 10.3 4.507 0.019 0.026 -0.007 0.044 less
Shrubland w/ PD 0.002 4.00 0.4 34.774 0.021 0.027 -0.006 0.049
Wooded/Riparian 0.057 7.00 10.7 1.295 0.037 0.036 0.002 0.073
Wooded/Riparian w/ PD 0.001 0.50 0.1 1.545 0.003 0.010 -0.007 0.012
Disturbed 0.067 12.00 12.5 0.017 0.064 0.046 0.018 0.110
Disturbed w/ PD 0.001 2.50 0.2 32.748 0.013 0.022 -0.008 0.035
Total 1.000 187.00 187.0 289.637 1.000
Z(l-a/2K) 2.576
1 = Z(l-a/2K) x (Pi(l-Pi)/n)l/2, K = number of habitat categories
2 = Observed use different from expected when 90 % confidence intervals around observed use falls below (less than expected)
or above (more than expected) proportion of habitat within the study area


and without the presence of prairie dogs showed that shortgrass with prairie dogs was
preferred and shortgrass without prairie dogs was avoided.
Home Range Results
Home range analysis using minimum convex polygons revealed that radio-
tagged ferruginous hawks occupied relatively distinct home ranges in 1991/92
(Figure 4.1). Home ranges of ferruginous hawks radio-tracked in 1992/93 showed
much more overlap than 1991/92 (Figure 4.2). In the winter of 1993/94, radio-
tracked ferruginous hawks again had relatively distinct home ranges (Figure 4.3).
Home ranges varied from 1.51 km to 58.13 km (Table 4.4). The average size and
the standard deviation (SD) of the home range of ferruginous hawks captured at
RMA decreased each year of the investigation (Table 4.4). Average home range was
29.5 km2 in 1991-92 (n = 6, SD = 23.9), 14.4 km2 in 1992-93 (n = 8, SD = 17.4), and
8.3 km2 in 1993-94 (n = 4, SD = 5.2). Wintering ferruginous hawks in 1992/93 also
had relatively smaller home ranges (Figure 4.2). The two individual hawks with the
largest home ranges appeared to concentrate activity in a relatively confined area for
a period of time, then shifted to another activity area.
The movements of tagged ferruginous hawks revealed several important
activity centers at RMA. Two major centers of diurnal activity occurred in the
northeastern and southwestern portions of RMA; two areas that supported large
74


75


tk
Figure 4.2 Home range of wintering ferruginous hawks radio-tracked on
Rocky Mountain Arsenal NWA in 1992-1993
76


'I t **f
1
c
2 MILES
l
3
2 KILOMETERS
Figure 4.3 Home range of wintering ferruginous hawks radio-tracked on
Rocky Mountain Arsenal NWA in 1993-1994
77


Table 4.4. Home range calculations using minimum convex polygons for radio-tagged ferruginous
hawks at Rocky Mountain Arsenal NWA, 1991-94
Bird # Area (km2) # of Locations
1991-92
1 58.13 64
2 13.04 32
3 7.47 110
4 57.24 39
5 33.95 50
6 7.33 76
Mean 29.53 61.83
SD 23.90 28.55
1992-93
7 9.08 146
8 1.58 17
9 28.69 198
10 2.27 27
11 4.71 205
12 50.97 200
13 1.51 33
14 15.76 54
15 2.38 20
Mean 18.24 102.40
SD 22.66 92.20
1993-94
16 3.45 33
17 4.92 60
18 9.91 41
19 14.86 129
Mean 8.29 65.75
SD 5.18 43.66
78


colonies of prairie dogs. The northwestern portion of RMA was also an important
center of activity for a few individual hawks.
Description of Communal Roosts
Radio-tagged hawks used a total of 29 different roost sites during the three
years of study. Sixteen of the 29 roosts were located at RMA. Most of these roosts
were used for a short period and by a single radio-tagged hawk; however, several
roost sites were used by numerous hawks for extended periods in all three years of
the investigation. Five important roost sites were identified on RMA based on the
frequency of use and the number of hawks using the roost (Figure 4.4).
Measurements were collected on four of these sites. I was unable to access the fifth
roost site located adjacent to a communal bald eagle roost due to USFWS
restrictions.
Wintering ferruginous hawks demonstrated strong roost site fidelity and
returned to the same roost 74 % of the time (Figure 4.5). Secondary roosts were used
21 % of the time, and all other roosts were used less then 5 % of the time.
All ferruginous hawk communal roost sites identified at RMA occurred in
isolated groups of trees that permitted access into the roost from all directions and
were generally adjacent to open grassland habitats. The number of trees in the roost
group ranged from 9 to 44 (Table 4.5). No apparent preference for tree species was
79


Figure 4.4 Locations of five main communal roosts used by ferruginous hawks
overwintering at Rocky Mountain Arsenal NWA 1991 1994
80


5%
% Time in main roost d% Time in secondary roost % Time in all other roosts '

Figure 4.5 Roost site fidelity of ferruginous hawks radio-tracked
on Rocky Mountain Arsenal N WR 1991-1994.
81


Table 4.5 Average measurements of ferruginous hawk communal roost characteristics -
Rocky Mountain Arsenal NWA, 1991 94
Roost # DBH (cm) Height (ft) Height (m) Proximity to Road Distance to P.D. (m) Dominant Tree
1 23.0 22.8 6.9 17.9 34.6 Elm
2 26.3 26.8 8.2 138.5 158.5 Elm
3 42.0 34.5 10.5 158.6 0.0 Cottonwood
4 123.7 66.8 20.4 160.2 0.0 Cottonwood
5 (unable to access roost # 5 due to Bald Eagle Management Area restrictions)
82


observed as both elms and cottonwoods were used equally. The average tree height
per roost ranged from 6.9m to 20.4m with an overall average height of 10.5m. The
average DBH per roost was highly variable and was directly associated with the
dominant tree species. Cottonwood dominated roosts were on average taller and
wider than elm roost trees. The proximity of roost sites to roads and prairie dog
towns was also highly variable at RMA (Table 4.5).
Discussion
Habitat Use
Two data collection techniques (roadside counts and radio-tracking) for the
analysis of habitat use and availability both showed that wintering ferruginous hawks
at RMA avoided crested wheatgrass, cropland and wetland habitats. In addition,
analysis of radio-tracking data also found that shrubland and weedy forb habitats
were avoided. This is contrary to analysis of roadside count data the showed that
weedy forb was the only habitat preferred by ferruginous hawks. One limitation of
the roadside count is that it only accounts for habitats within a restricted transect
width, which may or may not contain preferred habitats. In contrast, radio-tracking
has no spatial restrictions and, thus, can identify all habitats used by individual
hawks. The contrary results of the two techniques could be the effect of ferruginous
83


hawks selecting for another component of the habitat (i.e. prey) other than landscape
composition.
Analysis of habitat use and availability evaluated with the presence of prairie
dogs showed that wintering ferruginous hawks were strongly associated with prairie
dogs. Radio-tracking of individual hawks indicated that all habitats containing
prairie dogs were preferred and most habitats without prairie dogs were avoided.
Roadside counts also showed that shortgrass steppe with prairie dogs was the only
habitat preferred by wintering ferruginous hawks. This result is not surprising
considering that most prairie dogs occur in the shortgrass steppe. This habitat type
comprised about 10 percent of the landscape composition of RMA, and roadside
observation of ferruginous hawks in other habitats containing prairie dogs may have
been insufficient to be statistically significant. Prairie dogs provide a reliable prey
source and also modify habitats, creating burrows for other prey species such as
cottontail rabbits. This reliance, or selection of habitat associated with a single prey
species has been reported by several researchers. Lardy (1980) found that
ferruginous hawks selected habitats related to soil types corresponding to the hawks
principle prey species, the Townsend's ground squirrel (Spermophilus townsendii).
Howard and Wolfe (1976) suggested that shifts in vegetation type around successful
ferruginous hawk nests were related to jackrabbit densities.
84


Selection of habitats with prairie dogs may also be a function of prey
availability. Black-tailed prairie dogs keep vegetation surrounding their burrows
extremely low. This reduced plant cover can increase the detectability and
availability of all prey species. Previous studies on red-tailed hawks (Preston 1990)
and rough-legged hawks (Baker and Brooks 1981) suggest that foraging raptors
respond to the composite of prey density and plant cover density. Wakely (1978)
reported that ferruginous hawk foraging distribution was inversely related to plant
cover density.
Wooded/riparian habitat was the only habitat component that was selected by
ferruginous hawks with and without the presence of prairie dog. This is probably a
reflection of the preference of wooded areas for roost sites.
Other factors that may relate to habitat selection by wintering ferruginous
hawks at RMA include foraging behavior, detection of prey, perch availability, and
the presence or absence of competitors, particularly bald eagles. Foraging behavior
and, ultimately, habitat selection can be greatly influenced by perch availability
(Janes 1985). In a study of grassland raptors, Janes (1985) reported that ferruginous
hawks selected areas with relatively few perches and foraged more often from a low
soaring flight, while red-tailed hawks selected areas with high perch density and
foraged more often from perches.
85


I previously discussed some of advantages and disadvantages of the two data
collection techniques used in my study. Advantages of roadside counts include ease
of establishing and conducting surveys, low cost, use of basic field equipment, and
the ability to survey a population rather than a few individuals Disadvantages of
roadside counts include spatial and temporal restrictions, detectability, and the
influence of roads on raptor distribution. Radio-tracking has no spatial or temporal
restrictions and, based on my studies, may be able to provide a more accurate and
detailed analysis of habitat use. However, radio-tracking is labor intensive, involves
expensive electronic equipment, requires large sample size, requires more technical
skill to operate, may alter behavior, and is stressful to the target species. The use of
either of these techniques depends on the objectives of the study. Well funded
projects that require detailed accurate data, and have the opportunity to attach
transmitters to an adequate sample size should use radio-tracking techniques. Basic
low-budget monitoring studies that need general information or trend data should use
standard roadside counts.
Home Range
The reduction in size of ferruginous hawk home ranges corresponds with the
increase in average number of wintering ferruginous hawks observed per road
transect. Both these observations coincide with the increase in prairie dogs at RMA
(D. Serri, pers. commun.). This suggests that ferruginous hawks decreased the
86


average distance traveled in each year of the study, probably in response to the
increased prey availability at RMA and thus an increase in the area of the preferred
habitat type (shortgrass steppe with prairie dogs). By decreasing the distance
traveled to obtain necessary resources (food and cover), wintering ferruginous hawks
could reduce energy expenditures and increase survivability during this critical
period. Increasing density of ferruginous hawks in response to increased prey
confirms studies conducted by Schmutz and Hungle (1989) who found that the
density of breeding ferruginous hawks increase with ground squirrel abundance.
This indicates that when hawks are able to satisfy food requirements within a smaller
area, a higher density is tolerated. Thus, the increase in ferruginous hawks observed
per road count transect on RMA could be due either to an increase in the number and
density of hawks using RMA, or from ferruginous hawks constricting their home
range and spending more time at RMA, in both cases a response to increased prey
density.
Home range size estimation is subject to a researcher's choice of home range
model and is dependent on the method of data collection. Observations collected in a
short sampling period often are autocorrelated (Swihart and Slade 1985) which can
seriously bias results. The objectives of my study were to determine home-range
boundaries, specifically home-range of ferruginous hawks wintering at RMA. Short
sampling intervals are essential in this type of study and non-statistical estimates
87


(polygon measures) are recommended to minimize the effects of autocorrelation
(Swihart and Slade 1985).
I found that ferruginous hawk home ranges typically extended beyond the
boundaries of RMA. Frequently, hawks roosted at RMA, foraged throughout the day
off-site, then returned to roost at RMA. Thus, RMA may be regionally important to
wintering ferruginous hawks and conversely, off-site foraging areas, particularly
Stapleton International Airport and open fields in south Adams County, appear to be
essential or important resources to the management of ferruginous hawks at RMA.
This is especially pertinent in view of recent developments, including sylvatic plague
that has decimated RMA prairie dog populations, the closing of Stapleton
International Airport, and urban growth in Adams County that has eliminated many
of the foraging and off-site roosting areas described in this report.
Communal Roost
Communal roosts have been widely studied in bald eagles (Keister and
Anthony 1983, Grubb et al. 1989, Buehler et al. 1991) and common ravens (Engel et
al. 1992), but has received little attention for ferruginous hawks (Steenhof 1984). In
fact, few studies have been conducted on wintering ferruginous hawks in general,
which may account for the paucity of literature on communal roosting behavior.
88


Two aspects of ferruginous hawk communal roosting behavior were observed
at RMA that are similar to behaviors of other communally roosting species (Eiserer
1984). First, the same communal roosts were used year after year. Main communal
roosts, specifically in sections 5, 9, 11, and 25, were used in successive years by
ferruginous hawks wintering at RMA. At least three of these roosts were
consistently used by ferruginous hawks prior to the study (Beane pers. obs.).
Second, the number of birds at communal roosts fluctuated seasonally and yearly
(Eiserer 1984). The number of ferruginous hawks in a particular RMA roost varied
throughout the study and observations of radio-tagged hawks indicated that some
individual hawks moved between roosts. Although some movement between roosts
was observed, overall roost site fidelity by ferruginous hawks wintering at RMA was
strong. I frequently observed hawks flying past an established communal roost to
reside for the night in a more distant primary roost.
Although the adaptive significance of communal roosting is not well
understood, several hypothesis have been developed to explain this behavior. These
hypothesis include selection of microclimates that provide energetic benefits (Keister
et al 1985), and association with concentrated or sporadically abundant food
resources (Engel et al. 1992, Steenhof 1976). In addition to microclimate,
communal roosts may conserve energy as the raptors roost in the nearest roost site to
prey resources. Raptors, such as bald eagles, that are adapted to exploit seasonally
89


concentrated food resources (e.g., salmon) or depend on a single prey species may
be more likely to concentrate in communal roosts at seasonally or cyclically
abundant food resources. The adaptive significance of this behavior may include use
of the roost as information centers that alert roost members of available prey or allow
young to gain from the experience of adults in the roost (Ward and Zahavi 1972).
The ferruginous hawk in eastern Colorado is strongly associated with prairie dogs, a
species that historically and currently experience wide population fluctuations and
cyclic plague outbreaks. Considering this fluctuating prey base, it would be adaptive
for wintering ferruginous hawks to follow the crowd and roost near abundant prey
resources.
90


CHAPTER 5
OVERALL DISCUSSION AND CONCLUSIONS
Documenting relative abundance and distribution of raptors can detect trends
or determine general patterns of habitat use, but provides little insight into the
specific habitat components selected by diurnal raptors. Further investigation into
habitat use and availability based on landscape composition provides more
information on habitat selection. However, important components within habitats
may still be overlooked. Principal component analysis can help the researcher
ordinate species with respect to specific components, or habitat gestalt, within the
more general habitat classifications. My studies used a series of relative abundance
and distribution studies combined with habitat use and availability studies, and PCA
to provide a clear picture of diurnal raptor habitat use at RMA. These studies helped
me achieve my five primary objectives and reach the following conclusions:
1. raptors at RMA clearly indicate that this area is regionally important for raptors.
2. The distribution, abundance and habitat use of overwintering ferruginous The
distribution and abundance of both breeding and wintering diurnal hawks is
dependent on a combination of integrated habitat components consisting of; 1)
91


Full Text

PAGE 1

RAPTOR HABITAT RELATIONSHIPS AT THE ROCKY MOUNTAIN ARSE AL NATIONAL WILDLIFE AREA by Ronald D. Beane B.S. Colorado State University 1980 A thesis submitted to the University of Colorado at Denver in partial fulfillment of the requirements for the degree of Master of Arts Biology 1996

PAGE 2

This thes i s for the Master of Art s degree by Ronald D Beane has been approved Diana F. Tomback Jim K oehler D o u g las P Reagan ii

PAGE 3

Beane Ronald D. (M.A. Biology) Raptor Habitat Relationship s at the Rocky Mountain Arsenal ational Wildlife Area Thesis directed by Dr. Charles R. Preston ABSTRACT I investigated habitat use and distribut ion of diurnal raptors durin g a three years study at the Rocky Mountain Arsenal National Wildlife Area (RMA), northeast of Den ver, Colo r ado I used road side COLmts to document seasona l relati ve abundance and distribution of diurnal raptors and radio-trackin g studies to determine habitat use and general winter ecology of overwintering ferruginous hawks. Fourteen diurnal raptor species were identified at RMA. The most frequently encountered species were American kestrel (Falco sparverius) Swainson's hawk (buteo swa insoni ) and red tailed hawk (Buteo jamaicensis) in s ummer and bald eagle ( Haliaeetu s leucocephalus), ferruginous (Buteo rega li s), red-tailed and rough-legged hawk (Buteo lagopus) in winter. The mean number of r ap tors per summer survey was 21.0 in 1991 and 20 3 in 1992. Mean number of winter raptors observed increased from 24.0 in 1991-1992 to 30.7 in 1993-1994 due to increased numbers offerruginous hawk. The increase in ferruginous hawks was likely in response to increas e d prairi e dog populations. Winter distribution patterns confirmed that red-tailed hawks are associated w ith more trees than sympatric ha wks at RMA and ferruginous hawks are iii

PAGE 4

distributed in response to a prairie dogs. Diurnal raptors did at RMA not use habitat in proportion to availability. Weedy forb areas were preferred by summer American kestrels and winter ferruginous hawks. Wintering bald eagles rough-leg g ed hawks and red-tailed hawks preferred wooded/riparian and/or wetland habitats. Habitats avoided by most diurnal raptors in both summer and winter included crested wheatgrass and cropland. Ordination of species along habitat gradients (PCA) showed that red-tailed and Swainson s hawks occured in habitats with relatively lower diversity and more perches whereas American kestrels occurred in relatively di v erse habitats with few trees and perches. The presence of prairie dogs was the overriding factor in determining ferruginous hawk habitat use. All habitats with prairie dogs were preferred and most habitats without prairie dogs were avoided. Winter home ranges of ferruginous hawks became smaller in each successive year of the study ( 29.5 km2 in 1991-1992 14.4 km2 in 1992-1993 and 8.3 km2 in 1993-1994 ) This reduction in home range size coincided with increased abundance of prairie dogs. Overwintering ferruginous hawks showed strong fidelity to communal roosts This abstract accurately represents the content of the candidates thesis. I reccommend its publication Charles R. Preston lV

PAGE 5

v

PAGE 6

DEDICATION I would like to dedicat e this work to my lovin g wife, Janet and my children Chad and Kara who withstood countless days and hour s without a husband and father to provide them with guidance comfort and encouragement. Their patience and understanding is incredible. vi

PAGE 7

C O NTEN T S Ac kno w l e d g m e nt s .... ...... .... ........... ......... ... ...... ...... ... .... .... ... .... ......... .... .... . ix C H APTER 1 OV E RVI EW ..... ... ... ................. . .... .... .... .......... .... ... ....... ............ . ..... S tud y Area .... ......................... ... .................... ... ... ...... ........... ... ...... 5 2. RAP TOR AB UND ANCE AND DI ST RIB UT IO N A T RO CKY M O UNTAIN A R SENA L N A TIO NAL WI L D LIFE A REA Introduction ..... ........ .... .... .... ....... .... ...... ........... ... . ........... ... .... 8 Method s ....... .... .............. .... .... ........ ...... ...... ...... ... ...... ........ .. ... ... 9 R es ult s . . ...... ..... .... ..... ...... ....... .... ..... ... ........ ........ ....... ...... ... ... l3 Summer ...... .... ........ .......... ..... ...... ....... ............ ... ......... ......... 13 Wi nter ...... .... .... .... ...... ... .... ... .... .... ... ....... ........ ... . .... .... .... ....... 1 8 Mig r at i o n .... .................. ....... .... ........... .................................. ... 22 Dis c u ss i o n ...... ....... ... ........ .... ... ... ... ....... .... ...... .... ..... ...... ... .. ....... 23 3. HAB IT AT USE O F DIU RN AL RAP TORS AT R OC K Y M O UNTAIN ARSENAL NATI O NAL WI L D LIFE AREA I n t r od u ctio n .......... .................. . ...... ............. . .... ......... . .... ........ 34 Meth o d s ...... ... .... ............... . .... ......................... ............ ....... .......... 35 H a b itat Cl ass i fica tion ... ......... ... ......... ... .... ... ... ......... .... .... .... 35 R oa d si de C ount s ... .... ... ............. ......... ...... . .... ......... .... .... 3 8 Pri n c i p al C omp o n e nt Ana l ys i s ........... . ...... ... ............ ... .... ....... 39 R e sul ts .... .... .... ...... .......... ........... . ....... ....... .... .......... ....... . ..... .. 41 R es u lts of Habitat Use and Ava il a b i li ty fro Roa d s id e counts ........ .41 Pri n cipa l Co m po n e nt Ana l ys i s of R oa d s i de R a p tor Counts ........ .48 v i i

PAGE 8

Discussion .............. . ..................................... ............... ............ ....... 59 4. WINTER ECOLOGY OF FERRUGINOUS HAWKS AT ROCKY MOUNTAIN ARSENAL NATIONAL WILDLIFE AREA Introduction ... ....................... ..... .......... ................. ............... ..... ..... ..... 63 Methods .................... ... ...................... .................. .... ........................... 64 Habitat Use ........ ... .......... .......... ...... ........... ..... ............... .............. 64 Home Range of Wintering Ferruginous Hawks ......... .............. .... 66 Roost Site Characteristics ................. .................................... ........ 67 Results ..... ........ .... ... ... ...... ....... .... ...... ............. ... ........... ............... 68 Analysis of Habitat Use from Radio-tracking ........... ........... ....... 68 Analysis of Habitat Use from Roadside Counts ....................... ..... 70 Home Range Results ... ........... .......... .... .... ... .... ..... .... ............ ..... 74 Description of Communa l Roosts ... ........... ... ... ............................. 79 Discussion ..... ................................. ...................... ... .... ........ ... ..... ...... 83 Habitat Use ........ ........ ............... .... ........ ....... .... ....... ....................... 83 Home Range ..................... ....................... ..... .... ...... ............... .... 86 Communal Roosts . .... ............. ....... ..... ...................... ............. ... .... 88 5. OVERALL CONCLUSIONS AND DISCUSSION .............. ..... ........ ....... 91 REFERENCES CITED .................. ... .......... .... .......................... ... ................................. 94 VIII

PAGE 9

ACKNOWLEDGMENTS Funding for this project was provided by the US Army, under a cooperative agreement among the Army US Fish and Wildlife Service (USF WS) and the Den ver Museum of atural History ( DMNH). I would like to express deep g ratitude to m y graduate committee members Dr. D Tomback Dr. J. Koehler Dr. D. Reagan and in particular my graduate adviso r Dr. C. Preston who provided incomparable g uidance patience and encouragement. I wish to thank the entire staff of the Zoology Department at the DMNH Dr. R. Pieglar Dr. C. Jones Dr. C Meane y, and M. Castenenda. Personnel from the USFWS, especially D. Gober L. Malone M. Lockhart J.Griess and D Matiatos provided field assistance helpful suggestions and l ogistical support. D. Bocku s of the USFWS conducted most of the diet analysis and provided invaluable field assistance I am also thankful to C. Mackey Morrison Knudsen E n vironmenta l Services Inc. for providing insight into vegetation communities at RMA and to my field assistants particularly R. DeBaca and F. Hein who made data collection efficient rewarding and fun. lX

PAGE 10

CHAPTER 1 OVERVIEW In 1991 the Denver Museum ofNatural History (DMNH) began a series of ecological studies at the Rocky Mountain Arsenal National Wildlife Area (RMA) (Figure 1.1 ). These investigations evaluated the habitat associations of small birds s mall mammals, lagomorphs and raptors. The Rock y Mountain Arsenal ational Wildlife Area is currently on the ational Priorities List as a 'Superfund Site' and is undergoing environmental clean-up by the Army and Shell Oil Compan y (USFWS 1996) Current interim clean-up actions and eventual remediation of hazardous waste associated with RMA could potentially impact thousands of hectares of wildlife habitat. Wildlife management decisions and mitigation of environmental impacts of remediation will be implemented based on the best available information. Therefore, it is imperative to understand wildlife-habitat relationships including raptor-habitat relationships prior to remediation. Diurnal raptors are the most visible predators at RMA. Common diurnal raptors at RMA are red-tailed hawks (Buteo jamaicensis) Swainson's hawks (Buteo swainsoni), northern harrier (Circus cyaneus) and American kestrels (Falco sparverius) in summer; ferruginous hawks (Buteo regalis) red-tailed hawks rough legged hawks (Buteo lagopus), bald eagles (Haliaeetus leucocephalus) and golden 1

PAGE 11

BOULDER CO. I I I I 10 lo .(/) I i JEFFERSON CO. I o <-:>'j / I 25 Figure 1. I Study site location map WELD CO. ADAMS CO. ARAPAHOE CO. DOUGLAS CO. t N km j 2

PAGE 12

eagles in winter (Aquila chrysaetos). Because of their posit ion at the top of the food chain and their great mobility, raptors are valuable as environmental barometers (Craighead 1987 ). Diurnal raptors also control populations of herbivores (e .g., rodents, grasshoppers rabbits and hares) and are an important component of grassland ecosystems. In 1987 the National Wildlife Federation spo nsored a ser ies of raptor management symposia held regionally across the United States to discuss the status and importance of raptors. Both the Western and Southwestern Raptor Management Symposia identified winter habitat requirements of ferruginous hawks as an important future research need (LeFranc and Glinski 1988 Harlow and Bloom 1989). The ferruginous hawk (Buteo regalis) is a frequent overwintering raptor at the Arsenal. This species is endemic to grassland and open shrubland habitats in North America and is thought to be declining throughout much of its breeding range (Olendorff 1973 Houston and Bechard 1984 Schmutz 1984); howe ver, declines have only been confirmed in Utah and evada (Bechard and Schmutz 1995). Because of the high visibility and eco lo gical importance of diurnal raptors on RMA the U.S. Fish and Wildlife Service (USFWS) felt additional information was needed on raptor habitat relationships particularly habitat relationships of ferruginous hawk. The purpose of this study was to examine the temporal and spatial use of habitats by all common diurnal raptors and particularly to describe winter habitat requirements of ferruginous hawks. 3

PAGE 13

My objectives were to: 1 Document the year-round relative abundance and distribution of common falconiforms at RMA. 2. Determine seasonal habitat use of common diurnal raptors at RMA. 3. Determine winter habitat use of ferruginous hawks at RMA. 4. Document average home range of ferruginous hawks overwintering at RMA 5. Identify and characterize ferruginous hawk communal roost sites on RMA. I met my objecti ves with a combination of descripti ve studies and focused hypothesis testing. The primary focus of my study was to test the null hypotheses that 1) raptors use summer and winter habitat in direct proportion to its availability; and specifically 2) wintering ferruginous hawks use habitat in direct proportion to its availability. I used ancillary descriptive studies to document relative abundance and distribution of diurnal raptors in general and to specifically describe average home range and communal roost site characteristics of overwintering ferruginou s hawks. My study is divided into five separate chapters. This first chapter provide s a general introduction study organization, and study area description. The second chapter describes the relative abundance and distribution or diurnal raptors at RMA The third chapter builds on Chapter 2 describes seasonal habitat use and develops an ordination of coexisting diurnal raptors with respect to habitat components at RMA. 4

PAGE 14

The forth chapter focuses on the winter ecology of ferruginous hawks and builds on the general habitat use patterns described in Chapter 3. Chapter 5 provides overall summary and conclusions of my study. Detailed descriptions of methods used to meet my study objectives are provided in each chapter. Study Area My study area consisted of the 6800 ha Rocky Mountain Arsenal located 16 km northeast of downtown Denver Colorado (Figure 1.1 ). Originally shortgrass steppe dominated by blue grama (Bouteloua gracilis) and buffalograss (Buchloe dactyloides) (Lauenroth 1991 ), the RMA was converted to agriculture in the early 1900's. The area was purchased in 1942 b y the U.S. Army for the production of chemical weapons for World War II. After the war from 1947 to 1982 the Army, Shell Chemical and other chemical contractors manufactured chemical weapons and commercial pesticides at the installation. The area was listed as a federal Superfund site in the mid-1980's under the Comprehensive Environmental Reclamation Conservation and Liabilities Act (CERCLA) and is currently undergoing cleanup by the Army and Shell Oil Company. The Rocky Mountain Arsenal National Wildlife Refuge Act of 1992 (Public Law 102-402) established the RMA as a National Wildlife Refuge pending environmental cleanup Until then RMA is considered a National Wildlife Area. 5

PAGE 15

RMA is characterized by open expanses of disturbed and native grasslands. Disturbed grasslands are dominated by cheat grass (Anisantha tectorum ), kochia (Bassia sieversia na), field bindweed (Convolvulus arvensis) and planted areas of crested wheatgrass (Agro pyron cristatum). Native grasslands are dominated by sand dropseed (Sporobolus cryptandrus), needle and-thread grass (Stipa comata), and red threeawn (Ar istida lon giseta). Scattered among the open grasslands are shrub grassland areas dominated by yucca (Yucca glauca) and sand sagebrush (Artemisia jilifolia). The southern portion of my study area is characterized by grassland interspersed with woodland and riparian areas consisting of plains cottonwood (Populas sargentii), white poplar (Populas alba) locust (Robina spp .), and Chinese elm (Ulmus pumila). Three reservoirs (38 25 and 3.2 ha, respectively) and First Creek, an intermittent stream that flows from south to north through RMA, are the only reliable water sources within 1 0 km of the study area. The area surrounding RMA consists of cultivated cropland on the north and east, and urban development includin g Stapleton International Airport on the south and west. Although surrounded by urban development RMA remains a large island of open prairie habitat that supports di ve rse and abundant prey populations Currently the RMA is experiencing various degrees of habitat alteration as a result of interim response cleanup actions. These disturbances are anticipated to 6

PAGE 16

increase in the near future and continue for several decades as final remediation actions are implemented 7

PAGE 17

CHAPTER 2 RAPTOR ABUNDANCE AN D DIS T RIB UTION AT ROCKY MOUNTAIN ARSE AL NATIO AL WILDLIFE AREA Introduction Little is known about the distribution and abundance of raptors historically within the shortgrass steppe near Denver Colorado, particularly at RMA which was closed to the public for decades With the establishment of RMA as a ational Wildlife Area information was needed on wildlife habitat requirements to effectively manage the resource. Additionally information on fauna at RMA particularly top carnivores was needed to establish a pre-remediation bas eline. A study of the shortterm relative abundance and distribution of raptors at RMA can help determine trends of raptor occurrence within the boundaries of RMA and to some extent within the general region. This information can pro v ide resource managers with information to identify and manage high u se and low use areas on a species-specific basis. My objective for this study was to document the distribution and relative abundance of common diurnal raptors at RMA. 8

PAGE 18

Methods In order to determine relative abundance and distribution I conducted sixty roadside counts for all diurnal raptors for two consecutive years. These roadside counts were conducted semimonthly from June 1991 through June 1993 following the methods of Fuller and Mosher (1981, 1987). I conducted a dditional roadside counts in summer and fall 1993 to correspond with companion lagomorph studies. I conducted more counts in winter because the primary focus of this investigation was the distribution and relative abundance of wintering ferruginous hawks. I conducted roadside counts between 06:00 and 11:00 hours on days with favorable conditions for raptor observation (visibility exceeding 1500 meters and wind velocity less than 16 km/hour). My observations were conducted from a vehicle moving 25 to 40 km/hour, and all raptors observed within 400 meters of the road were recorded. During roadside counts I recorded start and finish odometer readings start and finish times location of bird (odometer reading Township / Range section, and map plot) distance to the bird (calculated from a map showing measured distances to perch sites), vegetation type surrounding the bird proximity to human disturbance bird behavior and perch type Data for all years were pooled by season for the analysis of raptor distribution. 9

PAGE 19

My raptor roadside count route was 39 km lon g consisting of maintained section roads within the boundaries of RMA (Figure 2.1 ) This route was subdivided into a series of 24 sample segments of approximately equal length ( 1.6 km). Starting and stopping points for each segment corresponded to section roads In order to maintain independence of observations, overlapping portions of sample segments were eliminated. Each segment was treated as an independent sample plot with specific habitat and land use characteristics. I divided the study area into five ecolog icall y similar regions that were representative of the vegetation topography land use soi l s and human presence within the study area (Figure 2 2). Regions consisted of 1) the northwestern region characterized by native grasses, cheatgrass, and crested wheatgrass; 2) the southwestern region characterized by scattered pockets of trees and windbreaks native grasses crested wheatgrass, and frequent human presence (passenger and private aircraft, heavy vehicle traffic administrative and maintenance buildings) ; 3) the southeastern region characterized by a mosaic of lak es canals wetlands wooded riparian areas and shrublands ; 4) the northeastern region characterized by open native grassland prairie dog colonies and a narrow riparian corridor along First Creek, and; 5) the central region characterized by historic and current human disturbance such as industrial and chemical facilities a dministrative buildings and areas of chemical contamination. 10

PAGE 20

/ $ 0 /? / /. 0 i/r ; ------------M 11n En11anc'l t;; w a: 1-"' ffi :> 0 J I -----; r; ,., 0 .., 0 .., .., .., .., 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1w w a: IV) 'J 00000000 I' Figure 2 1 Raptor roadside count transect route at Rocky Mountain Arsenal NWA Legend Established sect ion road -Transect route 11 t w w a: IV) ) ( ( 9th AVE 8thAVE 1w w a: IV) f N I ( :;; ( w ( a: ( :;; ( ( 0 ( I' ( ( ( ( ( ( ( ( c 0 0 0 0 ooooooO 0 I --7th liVE 6th AVE --

PAGE 21

4 SIXTH 1-lLI lLI a: EIGHTH 34 = 12 561h AVE Figure 2 2 Regions used for evaluation of raptor distribution and abundance on Rocky Mountain Arsenal NWA 12 (.')

PAGE 22

I divided counts into summer (breeding season) winter, and fall and spring migration. Following the methods of Fuller and Mosher (1987) I described species abundance and distribution within the RMA by recording the number of diurnal raptors observed within ecologically similar regions. I evaluated distribution and abundance within each region for the summer breeding season (May through August) and winter (November 8 through February) for all raptors that averaged more than one observation per roadside count. Because the five regions varied in size I standardized results by comparing the number of diurnal raptors observed per km of survey route within each region. Results Summer Eleven summer counts were conducted between 15 June 1991 and September 1992. Three species, red tailed hawk Swainson's hawk and American kestrel were observed nesting at RMA. Additionally non-breeding ferruginous hawks northern harriers and golden eagles were occasionally observed in summer (Table 2.1 ). The average total number ofraptors observed during summer roadside counts was 21.0 (SD = 2.35, n = 5) in 1991 and 20.3 (SD = 6.71, n = 6) in 1992 The American kestrel was the most abundant species observed in each summer (Figure 2.3) reaching peak abundance in late June and early July coinciding with fledging of 13

PAGE 23

Table 2 1 Diurnal raptor species occurrence by s eason at Rocky Mountain Arsenal National Wildlife Area Species Seaso n Winter Spring Summer Fall Red-tailed hawk X X X X Swainson s hawk X X X Ferruginous hawk X X X X Rough-le g ged hawk X X X C ooper s hawk X Northern harrier X X X X Bald eagle X X X Golden eag l e X X X X American kestrel X X X X Peregrine falcon X X X Prairie falcon X X X Merlin X Osprey X X Turkey vulture X X 14

PAGE 24

1991 12 00 IIFH RTH 10 00 IISH 8 00 6 00 IIAK 4 00 GE 2 00 0 00 MEAN 1992 -9 00 Oil FH 8 00 RTH 7 00 6 00 IISH 5 00 4 00 IIIAK 3 00 2 00 GE 1 00 0 00 MEAN Figure 2 3 Mean number of raptors by species observed on roadside counts co nduc ted on Rocky Mountain A r senal NWA Su m mer Le gen d : FH Ferruginous Hawk R T H Red-tailed Hawk SH Swainson s Hawk 15 NH Northern Harr ier AK American Kestre l GE Golden Eagle

PAGE 25

young. Both red-tailed hawks and Swainson's hawks reached peak abundance during August in both years as young fledged and became more visible. Analysis of summer distribution of raptors at RMA b y region revealed patterns for each of the three primary breeding raptors. American kestrels were most numerous in central and northeastern regions (Figure 2.4a). These regions were characterized by open landscapes dominated by weedy forbs and the central re g ion contained some of the most chemically polluted and physically disturbed areas on RMA Distribution of American kestrels varied somewhat between 1991 and 1992. In 1992 the average number of kestrels per km observed per transe ct decreased by 50 percent in the northeastern region and doubled in the northwestern and southwestern regions (Figure 2.4a). Red tailed hawks were unevenly distributed over RMA, occurrin g most often in the central region and rarely in the southeastern region (Figure 2.4b). Comparison of data between years showed fewer red tailed hawks were observed in 1992. The number of Swainson's hawks was also unevenl y distributed across regions of RMA. The greatest abundance occurred in the southwestern region (Figure 2.4c). I rarely observed Swainson's hawks in the southeastern re gio n Ferruginous hawks occurred sporadically in summer at RMA as vagrants or non-breeding subadults and were not observed to breed at the study area. I generally 16

PAGE 26

6 00 -,-----------------------------, 5 00+-----4 00 -t-----3 00 +-----2 00 1 00 0 00 NW SW SE 131991 AK I .1992 AK I Central NE J a. American Kestrel 0 3 0 2 0 2 0.1 0 1 0.0 I NW sw b. Red-tailed Hawk L 0 6 0 5 0.4 0 1 0.0 Hawk sw __ SE Central SE Central NE NE 0 1991 RTH RTH 01991 SH 2 SH Figure 2.4 Mean numb er of summer diurnal raptors per km observed during roads i de counts on Rocky Mountain Arsenal NWA. Data are presented by r egion and year 17

PAGE 27

observed ferruginous hawks in northern regions in summer especially the northeastern region. However the mean number of observations of ferruginous hawks was less than one per roadside count and insufficient for further evaluation. Winter I conducted twenty-five winter roadside counts between 15 November 1991 and 30 January 1994. Winter counts confirmed that the ferru g inous hawk redt ailed hawk rough-legged hawk (Buteo lagopus) and bald eagle (Haliaeetus leucocephalus) were common winter residents at RMA. Other species observed included golden eagle northern harrier, American kestrel merlin (Falco columbarius) prairie falcon (Falco mexicanus) Cooper s hawk (Accipiter coop e rii) and turkey vulture (Cathartes aura) (Table 2.1). The average number oftotal raptors observed per winter count increased from 24.0 (SD = 7.3 n = 11) in 1991-9 2 to 30.7 (SD = 6.5 n = 6 ) in 1993-94 primarily due to a notable increase in the average number of ferruginous hawks. Ferruginous hawks increased from an average of 8.6 (SD = 2.7 n = 11) hawks per count in 1991-92 to 11.5 (SD = 4.8 n = 8) and 12.2 (SD = 2.5 n = 6) hawks per count in 1992-93 and 1993-94 respectively The average number of red-tailed hawks observed was consistent over the three winters (5.9 6 .6, and 5.8) ; bald eagles and rough -l egged hawks varied slightl y from year to-year; golden eagles showed a steady increase over the three winters ; 1 8

PAGE 28

and northern harriers increased slightly in 1993-94 (Figure 2.5). Ferruginous hawk occurrence was greatest in the northeastern southwestern and central re gio ns w ith few hawks observed in the southeastern region (Figure 2.6a) Regional distribution was similar in all years but a slight decrease in ferruginous hawk abundance in southern regions was observed in 1993-94. This corresponded with an increase in abundance in the northwestern region (Figure 2.6a). Winter distribution of red-tailed hawks was fairly consistent over the three winters of observations. These birds occurred most frequently in the southwestern and southeastern regions (Figure 2.6b) Rough legged hawks occurred most frequently in the central southeastern and northeastern regions (Figure 2.6c ). Distribution of rough-legged hawks on a yearly basis was fairly consistent for the three winters of observations except that the average number of rough-legged hawks observed per roadside count doubled in the central region and decreased by 50 % in the southeastern region in 1992 (F igure 2.6c) The distributions of both bald eagles and go l den eagles were variable from year to year in winter at RMA. I occasionall y observed golden eagles during roadside counts primarily in the northeastern region. However, counts of golden eagles averaged l ess than one per roadside count and were not further evaluated Bald eagles occurred in the southeastern and northeastern regions and generally 19

PAGE 29

1991-92 1 0 00 8 00 6 00 4 00 2 00 0 00 MEAN 1992-93 15 00 1000 5 00 0 00 MEAN 1993-94 0 00 MEAN POOLED 12 00 10 00 8 00 6 00 4 00 2 00 0 00 FH RTH RLH NH AK BE GE II FH RTH IIRLH ONH AK OBE I OIGE II FH B RTH ONH AK OBE IJGE II FH RTH IIRLH IONH AK lOBE IIIGE iii FH RTH iii RLH ONH AK OBE IIGE Figure 2 5 Mean number of raptors by species observed on roadside counts conducted on Rocky Moun t a i n Arsenal NWA -Winter Legend : FH Ferruginous Hawk RTH Redtailed Hawk RLH Roughl egged Hawk AK Amer i can Kestrel BE Bald Eagle GE Golden Eagle 20 NH Northern Ha rr ier II

PAGE 30

0 50 0.40 CJ 1991-92 FH 0 30 -93 FH 0 20 0 1 993-94 FH 0.10 0 00 NW sw SE Ce n t ral NE a Ferruginous Hawks ----0 50 0.40 0 30 I CJ 19 91-92 RTH I 0 20 199293 RTH 0 10 0 1993-9 4 RTH N W SW SE Cen t ral NE d tailed Hawks --------------1 0.30 0 25 ----CJ 1 991-92 RLH 0 20 --0 15 1 992-93 RLH 0 10 0 1993-94 RLH 0 05 0 00 NW sw SE Central NE c. Rough-legged Hawks 0 50 0.40 CJ 1991-92 BE 0 30 -93 BE 0 20 0 1993-94 BE I 0 10 0.00 NW sw SE Central N E d. Bald Eagles Figure 2 6 Mean number of w intering diurn a l raptors per km observed during roadside count counts on Rocky Mountain Arsenal NWA Data are presented by region and year 21

PAGE 31

avoided central and northwestern regions (Figure 2.6d). I most often observed bald eagles in structurall y similar habitat consisting of large mature cottonwoods with open canopies and horizontal branches for perching. The southeastern and northeastern regions also contained bald eagle manipulation stations (artificial feeding stations) that tended to attract bald eagles to specific areas and accounts for some of the distribution patterns observed. Migration Twelve raptor roadside counts were conducted during fall migration periods (1 September8 ovember) ; seven in 1991 and five in 1992. Twelve roadsid e counts were also conducted during spring migration periods (1 March30 April); six each in 1992 and 1993 The average number of rap tors per fall transect was similar in 1991 (28.3 SD = 7.0, n = 7) and 1992 (29.2 SD = 6.7, n= 5). Spring road transect counts showed a marked increase in the average number of rap tors per transect in 1993 (24.3 SD = 9.4 n = 6) over 1992 (28.8 SD = 8.5 n = 6). Observations of rap tors during migration periods are highly variable and are dependent on weather patterns prevailing winds and the number of nonmigratory raptors present. This variability was apparent during roadside counts of several raptor species conducted during migration periods. The red-tailed hawk was the most frequently encountered raptor observed during fall migration in both years 22

PAGE 32

accounting for 36 5 %of observations in 1991 and 48.0 % of observations in 1992. The ferruginous hawk was the next most prevalent diurnal raptor during fall migration with about 21 %of observations in both years, followed by Swainson's hawk with 16.8 and 9.8 of observations in 1991 and 1992. Red tailed hawks per transect nearly doubled in spring 1993 (8.83, SD = 2.9, n = 6) from spring 1992 (4.83, SD = 4.8 n = 6) counts. American kestrels migrated through the study area in large numbers in spring 1992 accounting for nearly 50% of observations. However few kestrels occurred in spring 1993 Northern harriers were infrequentl y observed during both fall and spring migration periods. However harriers showed a bi g increase in average number per transect in spring 1993. Other species occasionally migrating through the study area included ospreys (Pandion haliaetus), peregrine falcons (Falco peregrinus), prairie falcons and go lden eagles Discu ssio n Frequently obser ve d diurnal raptors at RMA included red tailed ha wks, Swainson's hawks and American kestrels. American kestrel was the most frequently observed species. Frequently observed wintering raptors at RMA included ferruginous hawks red-tailed hawks rough-legged hawks and bald eagles. The ferruginous hawk increased in abundance each year of the study, while the abundance of red tailed hawk rough-legged hawk and bald eagle was relati ve l y 23

PAGE 33

stable The red-tailed hawk is the only species of raptor at RMA that is relatively abundant in both winter and summer. Detectability varies among raptor species according to body size, seasonal behavior habitat use and flight behavior (Fuller and Mosher 1981, 1987 ). This difference in detectability may have biased relative abundance estimates for diurnal raptors at RMA particularly for breeding American kestrels Observations of kestrels likely decreased as distance to the road increased due to their small bod y size (Millsap and LeFranc 1988). Red tailed and Swainson s hawks are relatively similar in size and it is assumed they are equally detectable in RMA habitats. The three major wintering buteos at RMA (ferruginous, red-tailed and rough -l egged hawks) are also relatively similar in body size and winter behavior and are assumed to be equally detectable in winter. I also minimized differences in detectability by standardizing the time of day seasons and weather of roadside counts (Fuller and Mosher 1987). Red-tailed hawks were most often observed during the breeding season in central regions containing scattered groves of large cottonwoods for perching and nesting and disturbed areas containing a high proportion of sparse vegetation and bare ground that may increase prey availability. Disturbed areas near waste basins and abandoned industrial complexes on RMA support a higher abundance of small mammals particul arly deer mice (Peromyscus maniculatus) than more native 24

PAGE 34

shortgrass habitats (MKE 1989 Boone and Preston 1994) I rarely observed red tailed hawks in the southeastern region although this region contained abundant riparian vegetation along lakes and canals However this region is also subject to fishing and wildlife viewing pressure along the road transect route. Red-tailed hawks may occur in areas further removed from human disturbance beyond the 400m transect width and thus be more abundant in the southeastern region than indicated by road transects. Although the typical width of roadside transects is 400 to 800 m wide (Kochert 1986 Millsap and LeFranc 1988) Millsap and LeFranc (1988) found that detectability of raptors decreased with increased distance and vegetation density. These researchers showed that detectability in summer deciduous woodland decreased sharply for models of three woodland hawks, including red-tailed hawk Based on these considerations some red-tailed hawks were undoubtedly undetected during summer roadside counts. However detectability is significantly higher in grassland than other vegetation types and consistent up to 300m (Millsap and LeFranc 1988). These authors also suggested adjustment of counts to account for detectability will improve accuracy but lower precision Furthermore unadjusted counts might be superior for monitoring studies where trends in raptor numbers over time is the objective ( Millsap and LeFranc 1988). Another consideration i s the general low abundance of summer red-tailed hawks at RMA. Typically, only one to 25

PAGE 35

three pairs of red-tailed hawks breed at RMA (ESE 1989 USF WS 1995 ) and a single nest near the roadside transect could influence observations and bias results Ferruginous hawks currently do not breed at RMA and no historical accounts of breeding exist although an occasional ferruginous hawk is observed in northern regions of RMA during summer. These northern regions are characterized by open shortgrass steppe and weedy forb habitats with large concentrations of pr air ie dogs The total number of ferruginous hawks observed was low (less than one observation per transect) and distribution could be hea vily influenced by a few indi vidua l hawks. The USFWS has installed artificial nest structures at RMA but they have not attracted any breeding ferruginous hawks The distribution of American kestrels was somewhat var iable bet ween years with a decrease of kestrels in the northeastern region coinciding with an increase in the southwestern and northwestern region. However American kestrel s in all regions appeared to be attracted to weedy areas. These areas ty pically support lar ge concentrations of grasshoppers and other insects at RMA ( Beane Unpubl. Data ) that provide prey for American kestrels (Sherrod 1978). Shifts in the summer distribution of breeding raptors could be caused by the relocation, abandonment, or establishment of one or more nest s ites or be a reflection of individual nest site success. Any successful nest site within the roadside count transect would increase the count within that specific region. An example of the influence of nest sites on 26

PAGE 36

summer distribution patterns of rap tors was e v ident in the v ariability of kestrel distribution between years and the notable decrease of Swainson's hawks in the southwestern region ofRMA in 1992. Two Swainson's hawk nest sites occurred within or near the transect corridor in this region in 1992 and the decreased abundance recorded during roadside counts probably resulted from decreased nest and fledging success between years. The diversity and abundance of wintering diurnal raptors at RMA indicate that the area is regionally important to wintering raptors. RMA supports one of the largest communal bald eagle roosts within the South Platte River Basin. Additionally, Christmas urban bird counts conducted by the Denver Audubon Society indicate that RMA supports more rough -le gged hawks bald eagles and golden eagles than other areas in the Denver-metro region ( Kingery 1996). The increase in wintering ferruginous hawks observed during road transects coincides with increased density of prairie dog populations (U SFWS 1994) and likely is a result of the expanded prey base attracting h awks to RMA rather than a trend in the wintering population of ferruginous hawks in the region. A review o f urban Denver Christmas bird count data from 1988 through 1995 showed a general decline in the number of ferruginous hawks observed within the Denver-metro region including the RMA. Some of this decline can be attributed to fluctuating prey populations. Prairie dogs at RMA were seve rel y reduced by an outbreak of 27

PAGE 37

sylvatic plague in 1988-89 and wintering ferruginous hawk population s declined accordingly. Some of the regional decline in 1989 90 was offset by healthy populations of prairie dogs and wintering ferruginous hawks at Buckle y Air National Guard Base (ANGB) (Beane Unpubl. Data Kingery 1996) Prairie dogs at Buckley ANGB experienced sylvatic plague in 1989 and regional ferruginous populations further declined (Kingery 1996). A recover y of prairie dogs at RMA through 1994 was reflected in a steady increase in populations of wintering ferruginous hawks, both at RMA and regionally (USFWS 1994 Kingery 1996). A second outbreak of plague at RMA in 1994 and continued l ow populations of prairie dogs at Buckley ANGB, combined with unprecedented growth along the Colorado front range that eliminated numerous prairie dog colonies grea tl y reduced counts of wintering ferruginous hawks in the Denver area. Wintering red-tailed hawks were associated with riparian habitat along First Creek in the southeastern and northeastern regions and woodland and shelterbelts in the southwestern region as well as abundant prairie dogs (Cynomys ludovicianus) and cottontail rabbits (Sylvilagus auduboni) for prey. These observation are consistent with studies that found red-tailed hawks associate more with trees than do sympatric ferruginous and rou g h-le gged hawks (Schne ll 1968 Jane s 198 5). Researcher s studying habitat partitioning between breedin g red-tailed ferruginous and Swainson's hawks have found that the three species exhibit a high de gree of 28

PAGE 38

dietary and habitat overlap and segregate primarily by partitioning nesting habitat (Cotrell 1981, Restani 1991) A comparison of summer and winter distribution of red-tails at RMA revealed a seasonal shift away from southern regions in winter to a more central distribution in summer (Figure 2. 7). This could be the result of seasonal changes in prey availabi lit y the availability of nest sites interactions with other nesting raptors or differences in detectability of raptors in wooded areas. However most wooded areas at RMA consist of relatively open and dispersed clumps of trees that permit good visibility within 400m of the road transect. Human presence and activity are known to alter raptor behavior (Sta lm aster and ewman 1978 Knight and Knight 198 4 Andersen et al. 1986 1989 1990 Holmes et al 1993). In this study diurnal raptors wintering at RMA did not avoid areas of heavy human pedestrian traffic in the central region of the arsenal. These results are consistent with Pre ston and Beane (1996) who using a different analysis (legit) found that raptors at RMA did not avoid heavily used roadways and may have habituated to relatively high levels of human activ it y found at RMA. A comparison of buteos wintering at RMA revealed severa l interesting distribution pattern s (Figure 2.8). Bald eagles were not included in this comparison because the practice of establishing eagle feeding stat ion s may bias their distribution. E qual numbers of red-t ai l ed hawks and ferruginous hawks were 2 9

PAGE 39

NW sw SE Central NE Figure 2 7 Comparison of summer and winter distr i bution of red-tailed hawks on Rocky Mountain Arsenal NWA 1991-1994 Legend : RTH-Red-tailed Hawk 30 CJ Winter RTH Summer RTH

PAGE 40

0 .70 0 .60 0 50 0.40 0 30 0 20 0 10 0 00 NW sw SE Cen tral NE IJFH RTH DRLH O Total Figure 2 8 Mean number of wintering buteos per km observed during roadside counts on Rocky Mountain Arsenal NWA Legend : FH Ferrug i nous Hawk RTH Red-tailed Hawk RLH Rough-legged Hawk 31

PAGE 41

observed in the southwestern region which accounted for this region supporting the highest number of buteos per km at RMA This region has also been identified as land to be auctioned or converted to cornrnerciaVindustrial uses as part of the RMA National Wildlife Area Act of 1992 (H.R. 1435). Two regions that also s upported a high abundance of ferruginous hawks (central and northeastern) had relatively low abundance of red-tailed hawks. Conversely the southeastern region supported a high relative abundance of red-tailed hawk s per km and th e lowest number of ferruginous hawks per km at RMA. Few buteos occurred in the northwestern region over the three years of observations. This region consists of large areas of disturbed and nati ve prairie and vast expanses of crested wheatgrass that has good potential for habitat enhancement for raptors. Some specific enhancement could include managing pre y populations particularly prairie dogs; providing habitat structure for alternate pre y, such as rabbits and mice ; remo v in g nesting perching and fora ging habitat from areas of contamination; providing poles and structures for perching, and trees and wind breaks for roosting and nesting in areas distant to contamination; pro vid ing visual shields to screen centers of human activity ; managing sylvatic plague, and ; coordinating with local county, and state governments to preserve and protect buffer areas and surrounding open space to maintain the ecological function of the RMA. Without buffer zones other nearby open space and movement corridors to connect 32

PAGE 42

RMA to surrounding habitat many of the summer and winter rap tor populations would s ub stantially d ec rease in number or disappear altogether. 33

PAGE 43

CHAPTER 3 HABITAT USE OF DIURNAL RAPTORS AT RO CKY MOUNTAIN ARSENAL NATIONAL WILDLIFE AREA Introduction Prairie ecosystems typically support numerous breeding and wintering raptors. Olendorff (1972, 1973) described the ecology and reproduction of raptor species at the Pawnee National Grasslands in northeastern Colorado but few studies have been conducted on raptors in the Denver-metro region. Habitat use or selection by diurnal raptors has been studied extensively (e g Jan es 1985, Craighead and Craighead 1956 Kimsey and Conley, Schmutz 1989, Per Widen 1994 ). Most of these studies on raptor hab it at use have focused on the reproduc ti ve period (but see Preston 1990) Objectives of this portion of my study were to: 1. Document year round habitat use and preference of all common diurnal raptors at RMA. 2. Develop an ordination of coexisting raptor species that describes how these species distribute themselves with respect to specific habitat components. 34

PAGE 44

In order to accomplish these objectives, I tested the null hypothesis that diurnal rap tors at RMA use habitat in direct proportion to its availability. Methods Habitat Classification Habitat is defined for this portion of my study as a spatially contiguous and primarily homogenous vegetation type that is distinctive from other such types (Partridge 1978 Hutto 1985) A landscape composition map (Figure 3.1) was generated from existing maps created on an Arc-info Geographic Information System (GIS) by Morrison-Knudson Environmental Inc. (MKE). Based on structural characteristics, vegetation map units delineated on the MKE map were combined into eight ecologically important vegetation types. These vegetation types were: Shortgrass steppe (including sand dropseed and needle-and thread grass) Shrub grassland (yucca and sand sagebrush ) Weedy forb Crested wheatgrass Wetland Woodland/Riparian Disturbed!Unvegetated Cropland (in areas adjacent to RMA) 35

PAGE 45

Figure 3.1 Rocky Mountain Arsenal lands cape composition IXSlUU!I!D liRlW3 I I WI!EDY !aUIS -MNilVIn!I'All
PAGE 46

An area was characterized as woodland if a minimum of five trees greater than 10 meters in height were in proximity (cano pies contiguous or nearly contiguous). Elms poplars and other non riparian tree species are depict ed as minor vegetation types on the map Cropland consisted of agricultural fields, primarily of winter wheat to the east of RMA. Cropland areas within 400 m of the eastern perimeter were included in roadside counts becau se they are occasionally used b y raptors. Seeded areas were classified as either shortgrass or disturbed depending on seed compositions and development stage. Open water habitats were assumed to be unavailable to raptors because these habitats were essentially frozen all winter and unused b y both winter and summer raptors I determined habitat availability b y using the GIS to calculate the area of each dominant vege tation type in square kilometers (km2 ) for the enti re st ud y area and each individual roadside count segment (approximate l y 1600 by 800 m). I then summed the percentage of each vegetation type available within each segment to obtain the percentage of habitat available over the entire roadside transect corridor. Each roadside count segment was annually groun d checked to ver ify vegetation classification and record changing landscapes (e .g. expanding prairie dog towns remediation, revegetation). I determined habitat use by locating each raptor on the 37

PAGE 47

landscape composition map (Figure 3.1) and recording dominant vegetation in a 100 m diameter circle surrounding each bird observed. The most prominent habitat type within the road count corridor was weedy forb which accounted for 36% of the landscape The next most prevalent habitat types consisted of shortgrass (20 %) and crested wheatgrass (19.5 %). The remaining five habitat types accounted for between 2 and 6 % of the landscape composition. For purposes of habitat use analysis each dominant vegetation component was considered a distinct habitat available to diurnal raptors. Roadside Counts Observations of diurnal raptors collected during roadside counts were summarized by the number of observations in specific habitat types. Habitat use availability analysis following the technique developed by Neu et al. (1974) and refined by Byers and Steinhorst (1984) was conducted for all road count observations. This procedure invol ves two steps. First a chi-square goodness of fit test of the hypothesis that rap tors use habitat in proportion to ava ilabilit y is performed Second if the null hypothesis is rejected, the preference or avoidance of each habitat type is determined based on Bonferroni z confidence intervals constructed around observed proportion of habitat use (Byers and Steinhorst 1984). 38

PAGE 48

In order to use the goodness of fit approximation the average (over all categories) of expected observations should be greater than six (Neu et al. 197 4 ). Principal Component Analysis After determining macro habitat selection of diurnal raptors at RMA based on landscape composition I used principal component analysis (PCA) to provide insight into how raptors distribute themselves with respect to habitat components. I separated road count data into summer and winter periods and analyzed 11 habitat variables potentiall y selected by the primary raptors seasonally occup ying RMA. Spring and fall migration periods were not analyzed due to yearly variability. The primary summer residents at RMA were American kestrel red-tailed hawk and Swainson's hawk. Primary winter residents at RMA were red-tailed hawk rough-legged hawk, ferruginous hawk and bald eagle. Habitat variables measured included five types of perch substrates (tree shrub telephone pole fence (8 ft. chain-link) and other human structures) five landscape composition types (c rested wheatgrass (CW) shortgrass steppe(SG) shrubland (SB), wooded/riparian(WD) and disturbed/bare ground (UB)) and distance to roads Distance to road was a perpendicular distance from the raptor to the nearest section road The following data were recorded for each raptor observed during road transects: 39

PAGE 49

Number of each perch type within 50 m radius of the raptor. Percentage of landscape composition type within a 50 m radiu s centered on the raptor (measured by mapping the UTM location of each raptor observed on the GIS vegetation map and delineating a 100 m diameter circle around each location). Distance to nearest section road in meters. Because of the difficulty in distinguishing between trees as a perch substrate and wooded/riparian as a vegetatio n cover type during PCA, I combined and simplified perch variables into a single perch var iable ( total number of perches within a 100 m diameter circle centered on the observed raptor). Percentages were converted using arcsin square root transformation tables Principal components analysis reduces the dimensions of a complex group of data by producing a smaller number of abstract variab l es This procedure ordinates species along axes that maximizes variance of the linear combinations of var iables. I conducted PCA of the correlation matrix of the reduced variable set usin g the PRINCOMP procedure in SAS (SAS Institute 1985) I then totaled and averaged scores of the principal components that accounted for the greatest variance in habitat use. I used Harvard Graphics (Bitstream Inc 1991) to plot two-dimensional plots of mean PCA scores plotted against each other. 40

PAGE 50

Results Analysis of Habitat Use and Availability From Roadside Counts I observed six species of diurnal raptors (bald eagle ferruginous hawk red tailed hawk rough-legged hawk Swainson s hawk, and American kestrel) frequently enough during roadside counts for analysis of habitat use and availabilit y Other diurnal raptors including prairie falcon peregrine falcon merlin golden eagle northern harrier turkey vulture and Cooper s hawk were observed on average less than once per roadside count and were not further e v aluated. Goodness of fit comparisons of roadside count data re v ealed that the actual number of all diurnal raptors observed within each habitat t y pe differed significantl y from the expected number of observations based on the occurrence of habitats w ithin the study area (X= 615.2 df= 7 P < 0.001). The null hypothesis was therefore rejected implying that diurnal raptors at RMA did not use habitats in proportion to availability. Confidence intervals revealed that raptors preferred (used more than expected) weedy forb wooded and wetland areas and avoided (used less than expected) all other habitat types (Table 3.1) To further understand raptor habitat use I anal y zed goodness of fit comparisons b y s eason on a species-specific basi s Summer breeding diurnal raptors at RMA consisted of primarily red-tailed hawk Swainson s hawk and America kestrel. Goodness of fit comparisons of 41

PAGE 51

Ta bl e 3 I Habitat distr ibuti o n of all diurnal raptor s at Rock y Mountain Arsenal NWA as recorded b y roadside count s total obse r vations 199194 Proportion Number of Pr oportio n of Confi denc e Int erval (90%) Observed Use Habitat of St ud y Number of Observation Chi-square Observations Lower U pper Differ ent From Type Area Observations Ex pected Contribution Pi zl Limit Limit Expected2 Cres ted Wheatgrass 0 195 157.5 252.34 35 643 0.122 0.023 0 099 0.145 le ss Weedy Forb 0 .362 543.5 467.56 1 2.336 0.420 0 034 0.3 86 0.455 more Short Grass 0. 201 204 259 .91 1 2 026 0 158 0 025 0.132 0 183 l ess Shrub land 0.057 42 73 95 13.806 0 032 0.012 0.020 0.045 less Wooded 0.023 112.5 29. 80 229. 514 0 087 0 020 0 067 0 .107 more Crop land 0.068 34 88 .50 33 565 0 026 0.011 0.015 0.037 less Ur ban/Di st urbed 0 059 45 75. 88 1 2.565 0.035 0 013 0.022 0.048 le ss Wetland 0.035 1 54.5 45. 07 265.739 0.11 9 0 023 0 097 0.142 more N Total 1.000 1 293 1293 6 1 5 194 1 000 Z( Ia/2 k) 2 5 I = Z(la/2K) x ( Pi ( I-Pi) / n ) 112, K = number of habitat categories 2 = Observed u se different from expected when 90% confidence int ervals around obse rved u se falls below (less than expected) or above (more th an expected) proportion of habitat within the study area

PAGE 52

roadside count data for these three species revealed that red-tailed and Swainson s hawks essentially used habitats in proportion to availability although Swainson's hawks used shrubland less than expected (Ta bles 3.2 and 3.3). However the reader should consider the results from red-tailed hawks with caution because the small number of expected observations ( 44) is less than an average of six over all categories and violates assumptions of goodness of fit tests. Summer kestrels used weedy forb habitats more than expected and shortgrass crested wheatgrass and cropland habitats less than expected (Table 3.4) Wintering raptors at RMA consisted primarily of bald eagles ferruginous rough-legged and red-tailed hawks. Golden eag l es, northern harriers and American kestrels also occurred in winter but not frequently enough for analysis Goodness of fit comparisons of roadside count data for the four primary wi ntering raptors revealed that none of these species used habitat s in proportion to availability. The number of bald eagles observed within each habitat type differed significantly from the expected number of observations based on the occurrence of habitats within the study area (X= 373.04 df= 7 P < 0.001). Bald eagles used wooded and wetland areas more than expected while using crested wheatgrass shortgrass shrubland, and urban/disturbed areas less than expected (Tab l e 3 5). Bald eag les were not observed in croplands. Wintering ferruginous hawks observed within each habitat type differed significantly from the expected number of observations (X= 53 7 df= 7 P 43

PAGE 53

Table 3.2 Habitat di s tribution of red-tailed hawk s at Ro c k y Moun t ain Arse nal NWA as r ecorded by roadsi de counts s um mer, 1991 93 Pr opo rtion Numbe r of Proportion of Co nfid e nce Int e r va l (90%) Observed Use Habitat of Study Number of Observation s Chi-square Observations Lower Uppe r Different From Type Area Observation Expected Contri bution Pi zl Limit Limit Ex p ec ted2 C rest e d Wheatgrass 0 .195 7 8 587 0 .2 93 0.159 0 1 38 0.021 0.297 Weedy Forb 0 3 6 2 22 15.911 2.331 0.500 0.188 0.312 0.6 88 Short Grass 0 201 6 8 84 5 0.915 0 1 36 0 1 29 0 007 0 266 Shrub land 0.057 5 .5 2.517 3.537 0 1 25 0.125 0.000 0 250 Wood e d 0.023 0 .5 1.014 0.261 0 .011 0 .040 0 029 0.051 Cropland 0.068 0 3 .012 3.0 12 0 000 0 .000 0.000 0.000 Urban/Distur bed 0.059 0 2 582 2.582 0.000 0 000 0.000 0.000 Wetland 0 035 3 1 .534 1.402 0 068 0.095 -0 .027 0 .163 Total 1 000 44 44 14. 332 1 000 Z(la/2 k ) 2. 5 I = Z( l -a/2 K ) x (Pi( 1-Pi) / n) 112, K = numbe r of habit a t categories 2 = Obse rv e d u se di fferen t from ex pect ed w h e n 90 %confidence interva l s around obse r ved u se falls below (le ss than expected) or above (more than expected) proportion of habitat w ithin th e study area blank spaces indicates habitat us e was in proporti o n to availability

PAGE 54

"""" '.JI Table 3.3 Habitat distribution of Swainson's hawks at Rocky Mountain Arsenal NWA as recorded b y roadside counts summer 1991 -93 Proportion Number of Proportion of Confidence Int erva l (90%) Observed Use Habitat of Study Numbe r of Observations Chi-square Observations Lower Upper Different From Typ e Area Observ a tion Expected Contribution Pi zl Limit Limit Exp ec ted2 Creste d Wheat g rass 0 .195 9 14.246 1 .932 0 123 0.096 0 027 0.219 Wee d y Forb 0.362 35.5 26 397 3. 1 39 0.486 0.146 0.340 0.633 Short Grass 0 .201 14.5 14. 674 0 002 0 199 0.117 0 08 2 0.315 Shrub land 0.057 I 4 .175 2.415 0.014 0 .034 -0 0 20 0 048 less Wooded 0.023 7 1.682 16. 807 0 096 0 086 0.010 0 182 C ropland 0.068 4 4.997 0 199 0 055 0 067 -0 .012 0 1 2 1 Urban / Disturbed 0.059 2 4.284 1.218 0 027 0.048 0.020 0 075 Wetland 0 035 0 2 544 2.544 0.000 0 000 0.000 0 000 Total 1 000 73 73 2 8 256 1 000 Z(l-a/2 k) 2 5 I = Z(l-a/2K) x (Pi(I-Pi) / n) 1 1 2 K = number of habit at categories 2 = Observed use different from expected when 90 % confidence intervals around obs erve d u se fall s below ( l ess than expected) or above (more tha n expected) pr o portion of habitat within the study area

PAGE 55

"""' 0\ Table 3.4 Habitat distribution of American kestrels at Rock y Mountain Arsenal NWA as r e corded b y roadside counts summer 1991 93 Proportion Number of Proportion of Confidence Interval (90%) Observed Use Habitat of Study Number of Observations Chi-square Observations Lower Upper Different From Type Area Observation Expected Contribution Pi z l Limit Limit Ex pected2 Crested Wheatgrass 0.195 11.5 20.687 4 080 0 108 0 076 0 033 0 .184 less Weedy Forb 0 3 62 64 38 330 17.191 0 604 0 119 0.485 0 .723 more Short G r ass 0 201 8.5 21.307 7.698 0 080 0 066 0 014 0 .146 less Shrubland 0 057 3 6 063 1.547 0.028 0.040 -0.012 0 069 Wooded 0.023 6 2.443 5.179 0 057 0 0 56 0 000 0 .113 Cropland 0.068 1.5 7.255 4 566 0 014 0.029 -0.015 0 043 less Urban/Disturbed 0 059 2 5 6.220 2 225 0.024 0.037 -0 013 0 060 Wetland 0.035 9 3.695 7 619 0.085 0.068 0 017 0.15 3 Total 1 000 106 106 50 .105 1 000 Z) l-a/2k) 2 5 I = Z(l-a/2K) x (Pi( l-Pi) /n) 1 1 2 K = numb e r of habitat categories 2 = Observed u se different from expected when 90 % confidence interv a l s around observed u se falls b e low (less than expected) or above ( more tha n expected) proportion of habitat within the study area

PAGE 56

Table 3.5 Habitat distribution of bald eagles at Rocky Mountain Arsenal NWA as recorded by roadside counts-winter 1991 94 Proportion Number of Proportion of Confidence Interval (90%) Observed Use Habitat of Study Number of Observations Chi-square Observations Lower Upper Different From Type Area Observations Expected Contribution Pi z Limit Limit Expected2 Crested Wheatgrass 0 195 14.83 2 12. 899 0 013 0.033 0 02 0 0 046 less Weedy Forb 0 362 19.5 27 48 2 2 318 0.25 7 0 .125 0 1 3 1 0 382 Short Grass 0.201 6 5 15. 277 5.043 0 086 0 080 0.005 0 166 less Shrub land 0 057 0 5 4 34 7 3.404 0 007 0 02 3 -0.01 7 0 030 less Wooded 0.023 24 1.752 282 603 0 316 0.133 0 182 0.449 more Cropland 0.068 0 5.202 5.202 0.000 0.000 0 000 0 000 .... Urban/Disturbed 0 059 0 5 4.460 3.516 0.007 0.023 -0.01 7 0.030 less -.] Wetland 0 035 24 2 649 172.098 0.316 0.133 0.182 0.449 more Total 1 000 76 7 6 487 083 1 000 Z(I-a/ 2k) 2 5 I = Z(l-a/2K) x (Pi(1-Pi) / n ) 112, K = number of habitat categories 2 = Observed use different from expected when 90 % confidence intervals around observed u s e fall s bel o w (le s s than exp ecte d) or above (more than expected) proportion of habitat within the study area

PAGE 57

< 0.001). The null hypothesis was therefore rejected implying that ferruginous hawks wintering at RMA did not use habitats in proportion to availability. Applying confidence intervals revealed that ferruginous hawks preferred weedy forb areas and avoided crested wheatgrass cropland and wetland areas ( Table 3 .6). The number of wintering rough-legged hawks observed also differed significant l y from the expected number of observations eX= 66 6 df= 7 P < 0.001). Rough-legged hawks preferred wetland areas avoided cropland and shortgrass steppe and used all other habitats in proportion to availability (Table 3.7). The number of wintering red tailed hawks observed also differed significantly from the expected number of observations (X= 515.9, df= 7 P < 0.001). Red-tailed hawks preferred wooded and wetland areas and avoided crested wheatgrass, weedy forb shortgrass shrubland and cropland areas (Table 3.8). Principal Components Analysis of Roadside Rap tor Counts Summer. The first two principal components accounted for 50.1 %of the total variance observed in summer roadside counts. The first principal component (PC 1 ) accounted for 25.76% of the variance (Tab l e 3.9). The highest positive correlation of the original variables (loading) is percentage of crested wheatgrass and the highest negative correlation is percentage of shortgrass steppe. This represents a gradient of vegetation diversity from more diverse shortgrass steppe to 4 8

PAGE 58

"""" \C) Table 3.6 Habitat distribution of ferruginous hawks at Rocky Mountain Arsenal NWA as recorded by roadside counts winter, 1991 94 Proportion Number of Proportion of Confidence Interval (90%) Observed Use Habitat of Study Number of Observations Chi-square Observations Lower Upper Different From Type Area Observations Ex p ected Contrib ution Pi z' Limit Limit Expected2 Crested Wheatgrass 0.195 16 36.494 11.509 0 086 0 051 0 .03 4 0.137 less Weedy Forb 0.362 109.5 67 620 25 938 0 586 0 .090 0.496 0.676 more Short Grass 0 .201 30 37 589 1 532 0 160 0 067 0 093 0 228 Shrubland 0 057 7.5 10.695 0.955 0.040 0.036 0.004 0.076 Wooded 0 023 5 5 4 310 0.329 0.029 0.03 1 -0 .001 0.060 C ropland 0 068 2 12.800 9.112 0 .011 0 019 -0 008 0 030 l ess Urban/Disturbed 0.059 14. 5 10.974 1.133 0 078 0 .049 0.029 0 1 26 Wetland 0.035 2 6.518 3.131 0.011 0.019 -0 008 0.030 les s Total 1 000 187 187 000 53.639 1 .000 Z(l-a/2k) 2 500 I = Z(l-a/2K) x (Pi( 1-Pi) / n) 112, K = number of habitat categories 2 = Observed u s e different from expected when 90 %confidence intervals around observed u se falls below (less than expected) or above (more than expected) proportion of habitat within the study area

PAGE 59

'JI 0 Table 3.7 Habitat distribution of rough-legged hawks at Rocky Mountain Arsenal NWA as recorded by roadside counts-winter 1991 94 Proportion Number of Proportion of Confidence Interval (90 % ) Observed Use Habitat of Study Number of Observations Chi-square Observations Lower Upper Different From Type Area Observations Expected Contribution Pi zl Limit Limit Expected2 Crested Wheatgrass 0.195 7 5 13. 856 2.916 0 106 0 .091 0 014 0 197 Weedy Forb 0 362 31.5 2 5 674 1 .322 0.444 0 147 0.296 0 .591 Short Grass 0 .201 7 5 14. 272 3.213 0 .106 0.091 0 0 1 4 0 .197 less Shrub land 0 057 4 4 .061 0 .001 0.056 0 068 -0 .012 0.125 Wooded 0.023 4.5 1.636 5 .012 0 06 3 0 072 -0 009 0 136 Cropland 0.068 0 5 4.860 3.91 1 0 007 0.025 -0 018 0 032 less Urban/Disturbed 0 059 2 4 166 1.127 0 028 0 049 -0 .021 0 077 Wetland 0 035 1 3.5 2.475 49 .122 0 190 0 116 0.074 0 307 more Total 71 71.000 66.6 2 3 1.000 Z(l-a/2k) 2.5 I = Z(l-a/2K) x (Pi(I-Pi) / n) 112, K = number of habitat categories 2 = Observed use different from expected when 90% co nfidenc e intervals around observed use fall s below (less than expected ) or above (more than expected) proportion of habitat within the study area

PAGE 60

Table 3.8 Habitat distribution of red-tailed hawk s at Rocky Mountain Arsenal NWA as r eco rd ed by roadside counts-w inter, 1991 94 Proportion Number of Proportion of Confide nce Int erval (90%) Observed Us e Habitat of Study Number of Observations Chi-square Observations Lowe r Upper Different From Type Area Observations Expected Contribution Pi zl Limit Limit Expected2 Crested Wheat g ras s 0.195 II 22 833 6 .133 0.094 0.067 0.027 0 1 6 1 less Weedy Forb 0 362 26 42. 3 08 6 286 0.22 2 0.096 0.126 0.318 les s Short Grass 0.201 12.5 23.518 5.162 0.107 0.071 0.035 0 178 less Shrub land 0.057 1.5 6.692 4.028 0.013 0.026 -0.013 0.039 less Wooded 0 023 33 2 .696 340.558 0 .2 8 2 0.104 0.178 0.386 more C ropland 0.068 1.5 8 .008 5 2 89 0 013 0 026 -0 0 1 3 0 039 les s 'Jl Urban/Disturbed 0.059 3 6 866 2.177 0.026 0.037 -0 0 II 0.062 ...... Wetland 0 035 2 8.5 4 078 146 261 0 244 0.099 0. 144 0.343 m o re Tota l 1 000 I 17 117 5 1 5 894 1 000 Z( l-a/2 k) 2.5 I = Z(l-a/2K) x (Pi(l-Pi)/ n) 112, K = number of habit a t categories 2 = O b se rv e d use di ffe r ent from expected when 90 % co nfid e n ce inte rval s around observed u se fall s below ( l ess than expected) or a bov e (mor e th an ex p ec t e d ) proportion of habitat within th e study a rea

PAGE 61

Table 3 9 Sum m ary sta ti stics of a prin cipal compon ents analysis of diurnal raptor summer habitat components on Rock y Mountain Arsenal NW A Principal comp onents Statistic 2 3 Eigenvalue 1.8 1.7 1.12 %of variance 25 .76 24 34 16. 05 Cumulative% 25.76 50 1 66.15 Component Eigenvectors Perch 0.265 0 599 0.107 cw 0.485 -0.457 -0.293 SG -0.694 0 .161 -0 162 SB 0.084 -0.049 0.930 WD 0.268 0 630 -0.006 UB 0.205 -0 052 -0 104 RD 0 303 0 067 0 009 4 1.05 1 5 .01 81.16 -0.154 -0 334 0 082 -0.019 -0 089 0 844 0.371 Perch =Number of p erches (tree s, pol es, structures) w ith in 1OOm diameter circle of observed rap tor CW = crested wheatgrass SG = shortgrass steppe, SB = shrubl and WD = Wooded/Riparian UB = Urba n f un vegetated. RD = Distance t o neare st road 52

PAGE 62

monotypic crested wheatgrass habitat. Species found in more monotypic and introduced habitats would be expected to have high PC 1 values. Species found m more diverse habitats would be expected to have low PC 1 va lues The second principal component (PC 2) accounted for an additional24.34% of the variance observed in s ummer roadside counts (Table 3.9) The highest positive loadings are the number of perches and the percentage of wooded/riparian habitat. The highest negative loading is the percentage of crested wheatgrass habitat. This component represents increasing woodland and number of perches. Species found in wooded areas with numerous perch sites would be expected to have high PC 2 values Species found in habitats with little woodland / riparian vegetation and few perches would be expected to have low PC 2 values Habitat relationships of diurnal raptors on RMA can be reconstructed by using the principal components as coordinates in a two-dimensional plot (James 1971 ) Figure 3.2 presents the relationship of summer diurnal raptors along the axis of the first two principal components. The horizontal axis representing PC 1 has separated the species along a vegetation diversity gradient. Swainson s hawk occurred in more monotypic crested wheatgrass habitat than red-tailed hawk and especially American kestrel which occurred in more diverse habitats (Figure 3 .2). The vertical axis representing PC 2 separated the species along a gradient of increasing perches and wooded/riparian habitat. Red-tailed hawks occurred in 53

PAGE 63

Wooded/Riparian 0.3 Perch es C\1 (.) a. 1\ "'C c:: "'C 0 0 3: -"' vegetation Vegetation diversity 1 Figure 3 2 Two-dimensional ordination of three species of summer diurnal raptors at R ocky Mountain Arsenal WA acc0rding to the first two principal co mpon en t s ( PC I and PC 2) Locations of rap tors along plot axes repres ents mean principal compo n ent scores. AK =American kestrel RTH = Red-iailed hawk, SH = Swainson's hawk

PAGE 64

habitats with more perches and woodland than either kestrels or Swainson s hawks American kestrels occurred in areas with the fewest number of perches and little woodland habitat (Figure 3 .2) Winter. The first two principal components accounted for 49.0 % of the total variance observed during winter roadside counts The first principal component accounted for 29.1 %of the variance (Table 3.10) The highest positive loadings are the percentage of wooded / riparian habitat and number of perches. The only negative loading is the percentage of shortgrass steppe habitat. This represents a gradient of increasing wooded/riparian habitat and perch availability and decreasing grassland. Species occurring in more wooded habitats with numerous perches would be expected to have high mean PC 1 values Species occurring in grassland habitats with few perches would be expected to have low mean PC 1 values. The second principal component accounted for an additional19.9 % of the variance observed during winter roadside counts The highest positive lo a dings are percentage of shortgrass steppe and wooded/riparian habitats. The most negative loadings are percentage of crested wheatgrass and urban/unvegetated habitats ( Table 3 10) This represents a gradient ofveg etation cover and div ersity from disturbed bare ground habitats to more native and diverse habitats. Species occurring in diverse native habitats would be expected to have high PC 2 v alues Species 55

PAGE 65

Table 3.10 Summary statistics of a principal components analysis of diurnal raptor overwintering habitat components on Rocky Mountain Arsenal, NWA Principal components Statistic I 2 3 Eigenvalue 2 .04 1.39 1.08 % of variance 29 .13 19.89 15.39 Cumulative% 29.13 49.02 64.41 Component Eigenvectors Perch 0.516 0.261 -0.167 cw 0 .101 -0.749 -0. 300 SG -0.59 5 0.393 -0.143 SB 0 .261 0 016 -0.084 WD 0 .548 0 .317 0.070 UB 0 015 -0.263 0 .848 RD 0 004 0 .215 0 .361 4 1.04 14.82 79' .23 -0. 094 0.087 -0. 019 0.702 0 282 -0.162 0 620 Perch = Number of perches (trees, poles structures) within I OOm diamet e r circle of obse rved raptor CW = crested wheatgrass, SG = s hortgra ss steppe, SB = shrubland WD = Wooded/Riparian UB = Urban / unvegetated RD = Distance to neare st road 56

PAGE 66

occurring in monotypic crested wheatgrass and disturbed unvegetated habitats would be expected to have low PC 2 values. Figure 3 3 presents the relationship of winter diurnal raptors along the axis of the first two principal components The horizontal axis representing PC 1, has separated the species along a gradient of increasing woodland and number of perches. The bald eagle and red tailed hawk produced high mean value PC 1 scores indicating that these spec ies occurred in wooded habitats with a high de gree of perch availability. In contrast the ferruginous h awk produced the lowest mean PC 1 score indicating that this species occurred in grassland habitats with a low degree of perch availability. Rough-legged hawks occurred in grasslands with relativel y more woodlands and perches than ferruginous hawks but with notably fewer woodlands or perches than either red-tailed hawks or bald eagles. The vertical axis representing PC 2 separated wintering raptors at RMA along a vege tation cover and diversity gradient (Figure 3.3) T he bald eagle produced the highe st mean PC 2 score indicating that this spec ies occurred in much more diverse habitat than ferruginous, red-tailed and rough-legged hawks. The ferruginous hawk produced the lowest mean PC 2 score indicating that this species occurred in habitats with relatively lower vegetation cover and diversity (Figure 3.3). 57

PAGE 67

'.Jl 00 Diverse 0.8 vegetation BE 1\ 0.6 ... . . . ... . I ........ >. -(I) 0.4 .... ...... > "0 C\1 "0 c:: u "' 0.2 .... a. I > 0 u 'lH c:: 0 Q RLH "' Q) 00 + -0.2 FH disturbed -0.4 vegetation -1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 PC 1 Perches Grasslands ------------------------------> Wooded I Riparian Increasing woodland Figure 3.3 Two-dimensional ordination of four species of winter diurnal raptors at Rocky Mountain Arsenal NWA according to the first two pr. incipal components (PC 1 and PC 2) Locations of raptors along plot axes represents mean principal component scores. AK =American kestrel, RTH =Red-tailed hawk, SH = Swainson's hawk

PAGE 68

Discussion The winter and swnmer distribution of raptors at RMA are undoubtedl y related to habitat selection. I found that breeding diurnal raptors at RMA segregated in relation to habitat. American kestrel occurred in grasslands red-tailed hawk in more wooded areas with numerous perches and Swainson s hawk occupied habitats in-between. Schmutz et al. (1980) in a study of sympatric buteos breeding in grassland habitats in Canada found a similar separation among species, with red tailed hawks associated with wooded parkland; ferr uginou s hawks with open prairie ; and Swainson's hawk occupying the ecotone between wooded parkland and open prairie. Janes (1985) in a similar study found red-tailed hawks in areas associated with relatively high perch density ferruginous hawks in areas of relativel y flat w1broken habitat without perches and Swainson's hawks in relatively flat areas containing a few widespread perches Smith and Murphy ( 1973) found similar habitat relationships in U tah and also found that topographic relief and specific plant associations were important to hawk distribution. Swnmer and winter raptors at RMA se l ect habitats disproportionate to avai labili ty. Swnmer American kestrel preferred weedy areas while avoiding shortgrass and cropland areas. This finding is consistent with distribution patterns described in Chapter 2. Summer red-tailed hawks and Swain son's hawks generally 59

PAGE 69

used habitats in proportion to availability but Swainson s hawks avoided shrubland areas. Although shrubland areas support some of the highest densities of small mammals at RMA (Boone and Preston 1995 ) the structure and height of the vegetation may make prey unavailable to buteos ( Pre ston 1990 Wakely 1978 Baker and Brooks 1981, Bechard 1982) Integrating raptor distribution results with habitat use and availability results provides a clearer picture of winter rap tor ecology at RMA. Wintering ferruginous hawks prefer weedy forb habitats primarily in the northeastern and central regions of RMA. Red-tailed hawks prefer wooded and wetland habitats primarily in the southwestern and southeastern regions of RMA which contains the highest percentage of these habitats. Bald eagles also prefer wooded and wetland habitats primarily associated with the lakes and First Creek in the southeastern and northeastern regions of RMA. Measurements from roadside counts are organized by in di vi dual species observed with little attention paid to other nearby species. This approach assumes that predictable relationship s exist between the occwTence of a bird and its characteristic vegetation and/or habitat requirements, or niche-gestalt (James 1971). Inherent to the term gesta lt are th e concepts that each species has a set of proximate factors that it responds to and that it has a predetermined set of specific search images (Tinbergen 1951 ). The habitat gestalt for summer species of diurnal raptors 60

PAGE 70

at RMA determined by combining PCA and habitat use / availability analysis provides a clear picture of breeding raptor habitat gestalts. American kestrels occurred in diverse vegetation with few perches particularly weedy forb areas that typicall y support abundant insect populations. Red-tailed hawks most frequently occurred in disturbed centra l regions of RMA associated with non-native vegetation and adequate trees for perching. Swainson's hawks also occurred in monotypic non native vegetation but with fewer perches and woodland areas than red-tailed hawks. The gestalt for wintering raptors at RMA was fairly evident for the four common species Both PCA and habitat use and availability analysis indicated a gestalt of grassland habitats generally lackin g perches for wintering ferruginous hawks at RMA. These results confirm the findings of Janes (1985) and Bechard and Schmutz (1995). The gestalt for both bald eagles and red tailed hawks wintering at RMA was wooded habitats with numerous perches. However bald eagles occurred in areas with greater diversity and vegetation cover. Rough-legged hawks appeared to use most habitat types although they tended to prefer wetland areas and avoid shortgrass and cropland areas. Summer American kestrels and winter ferruginous hawks both preferred weedy forb habitat. However kestre l s used habitats with high vegetation cover and diversity ; whereas wintering ferruginous hawks used areas with low vegetation cover and diversity. This apparent contradiction can be explained by the species selecting for 61

PAGE 71

specific habitat characteristics within the more general weedy forb type American kestrels selected weedy forb areas that contained diverse mosaics of weedy forbs and grasslands. These areas support high concentrations of prey (e.g., grasshoppers and grassland birds) in summer that likely attract American kestrels. However wintering ferruginous hawks selected low-growing weedy areas with little diversity specifically prairie dog towns. Although PCA and habitat use / availability analysis provided a habitat gestalt for summer and winter raptors at RMA, other community factors, such as competition and prey availability may have been missed in the analyses that could also influence habitat use (see James 1971). Vegetation structure has also been associated with habitat selection and distribution in raptors and was not specifically investigated during this study. However Janes (1985) and Preston (1990) found that vegetation structure alone did not exert a strong influence on habitat occupied by hawks. Prey related habitat selection has also been suggested for red tailed and rough-legged hawks (Schnell 1968) and ferruginous hawks ( Howard and Wolfe 1976, Lardy 1980). Prey abundance and availability components could have strengthened my analysis, particularly for summer data. Several studies (Preston 1990 Baker and Brooks 1981, Schmutz and Hungle 1989) have found that raptor distribution and territory size is influenced by prey. This component may be critical for breeding raptors that need abundant and available prey to feed young. 62

PAGE 72

CHAPTER4 WINTER ECOLOGY OF FERRUGINOUS HAWKS AT ROCKY MOUNTAIN ARSENAL NATIONAL WILDLIFE AREA Introduction The ferruginous hawk is currently classified as a candidate species for the federal Threatened and Endangered Species List under the Endangered Species Act of 1973 (16 U.S.C.). The primary reason for considering the listing is a continued decline in populations resulting from habitat destruction and reduced prey populations (caused by agricultural practices ) (USFWS 1991). Olendorff(1993 ) confirmed population declines only in Utah and Nevada and attributed these declines primarily to cultivation. Grazing small mammal control programs, mining and fire also contribute to ferruginous hawk declines (Olendorff 1993). Recent emphasis on conservation of wintering grounds and migration corridors for migrant species ( Terborgh 1992) has underscored the importance of identifying essential habitat components in all ecosystems in which a species occurs. As a migratory raptor the overwintering condition of ferruginous hawks is important to the overall reproductive rate ofthe species. The goal of this portion of my study was to identify and describe several 63

PAGE 73

ecological aspects of overwintering ferruginous hawks at RMA, including detailed habitat use home range and characteristics of communal roosts. Although numerous investigators ha ve described the habitat use of breeding ferruginous hawks and other sy mpatric prairie species of raptors (Schmutz et al. 1980 Cotrell 1981, Bechard et al. 1990 Restani 1991 ) little information is available on the habitat use and home range of wintering ferruginous hawks. Communal roost use b y wintering ferruginous hawks has also rarely been reported in the literature (but see Steenhof 1984). Ferruginous hawk communal roost sites first identified at RMA by M. Lockhart (USFWS) in 1988 are further described in thi s portion of my study Methods Habitat Use Habitat use b y wintering ferrug inou s hawks at RMA was analyzed using data collected by two different methods; roadside counts and radio-tracking. I analyzed habitat use and availability from both radio-tracking and roadsi d e count data according to the techniques ofNeu et al. (1974) and Byers and Steinhorst (1984) as described in Chapter 3. RadioTracking Twenty-eight ferruginous hawks were trapped at RMA using the modified Lockhart method (Bloom 1987) durin g three consecutive winters from 1991 through 1994. I attached tail-mounted radio transmitters to the central 64

PAGE 74

rectrix of captured hawks and secured the transmitter and antenna with thread and epoxy (Kenward 1987). Sufficient data to analyze habitat use / availability were collected for 17 radio-tagged hawks. I summarized the total amount of time ferruginous hawks were observed in each respective habitat type. Analyses were restricted to observations of radio-tagged hawks within an area that consisted of RMA and a buffer area that included adjacent portions of Stapleton International Airport (Sections 9 and 1 0) and a 400m corridor along the eastern boundary. This buffer zone corresponded to areas covered b y the roadside count transect. The habitat available to hawks was assumed to consist of all habitats within RMA and the buffer area For purposes of habitat use analysis each dominant vegetation component was considered a distinct habitat available to ferruginous hawks. Prairie dog distribution was calculated based on maps and estimates compiled by the USFWS on a yearly basis. The distribution of prairie dogs within each habitat component consisted of the average extent of prairie dogs over the three years of study. Roadside Counts. I observed ferruginous hawks during roadside counts as described in the previous chapter for diurnal raptors Observations of fly ing ferruginous hawks were included in relative abundance estimates described in Chapter 2, but were excluded from habitat use and availability analysis 65

PAGE 75

Home Range of Wintering Ferruginous Hawks Radio-tagged ferruginous hawks were tracked during fall and winter 1991-92 1992-93 and 1993-94 During the first winter of study radio-tagged birds were tracked for two 4-hour periods per week (systematic relocations) (Andersen and Rongstad 1989). During the final two winters of study radio-tagged birds were tracked for a 12-hour (dawn to dusk) period every 5-7 days. In order to minimize bias caused by capture stress, I began tracking radio-tagged hawks two to five days after capture (White and Garrot 1990). I then tracked hawks systematically until the individual hawks either shed the transmitter or left the area and could not be relocated. Hawk relocations were marked on a 1 :24,000 topographic maps and recorded by Universal Trans-Mercator (UT M) to the nearest 10 meters. Additionally, as many radio-tagged birds as possible were visually relocated once per day two to five days per week (random relocations) to help determine spatial distribution patterns of ferruginous hawks at RMA. Home range size for each radio-tagged ferruginous hawk was estimated using the minimum convex polygon approach (Mohr 194 7 Southwood 1966). Areas of high diurnal ferruginous hawk use were identified using the 50% harmonic mean activity area (Dixon and Chapman 1980) Home range analyses were conducted on 66

PAGE 76

McPaal (Stuwe 1992) home range computer software. All non-identical consecutive relocations were considered data points for home range analysis. Roost Site Characteristics Communal roosts were defined as a single tree or a group of trees containing more than four ferruginous hawks for a period greater than a week. Be ginning in November 1992 observations of known ferruginous hawk communal roost sites were conducted during dusk at least once every week to determine hawk use and density at communal roosts. I also conducted dusk surveys for unknown roost sites once every month. Communal roost sites identified by radio-tracking and dusk surveys were mapped photographed and studied according to methods adapted from Keister and Anthony (1983). Roost trees were measured after wintering hawks had left communal roost s for the season and before s ummer bre edi ng raptors began nesting. Data collected from communal roosts included: 1) tree species and number of trees in the roost group 2) diameter at breast hei g ht (DBH) 3) height classification (0-15, 16-23m 24 -30m ), 4) structure classification 5) number of roost trees within a communa l roost 6) protection from prevailing wind 7) proximit y to roads and other man-made structures, and 67

PAGE 77

8) distance to prairie dog colonies. Roost site fidelity was determined from radio-tracking of individual ferruginous hawks. Fidelity was calculated based on a simple percentage of nights observed in individual roost sites. Results Analysis Of Habitat Use From Radio-Tracking Goodness of fit comparisons of radi o trac king data revealed that l ocations of ferruginous hawks at RMA differed significantly from the occurrence of habitat types within the study area (X= 132.7 df= 7 P < 0.001 ) (Ta ble 4.1). The r efore the null h yp othesi s that ferruginous hawks used habitats in proportion to ava ilabili ty was rejected Confidence intervals indicated that radio-ta gged ferruginous hawks used shortgrass wooded and urban/disturbed areas more than expected and used all other habitats less than expected. Habitats were then combin e d into structurally s imil ar habitat components to d etermine if hawks s ho wed simi lar h abitat se l ect i on based primarily on vegeta tion structure versus composition. T he vegetation classes combined into s ingle habitat components included the follo w ing: Wooded-Wetland combined into a single wooded / riparian component Crested wheatgrass Cropland combined into crested wheatgrass component (t hes e two habitats are s tructurall y s imilar in winter and 68

PAGE 78

0\ \0 Table 4.1 Habitat distribution of overwintering ferrugin o u s hawk s ( n = 17) a t Rocky Mountain Arsenal NWA as recorded by radio-tracking total observations 1991 94 Proportion Proportion Confidence Interval (90%) Habitat of Study Minutes of Total Minutes Chi-square Lower Upper Type Area Observed Observations (Pi) Expected Contribution Pi zl Limit Limit Crested Wheatgrass 0.191 1114 00 0 .031 6833.19 13. 375 0.031 0.002 0.029 0.033 Weedy Forb 0 .33 6 10466 50 0 29 2 12035.45 0 .571 0.292 0 006 0.286 0.298 Short Grass 0 240 1 35 49 00 0.379 8601 84 7.950 0 379 0 006 0.372 0.385 Shrub land 0 063 323. 50 0.009 2255.25 4 623 0 009 0 .001 0.008 0.010 Wooded 0.024 6434.50 0 180 849 .61 102 577 0 180 0.005 0.175 0.185 C ropland 0 028 360. 50 0 010 990 70 1. 120 0 010 0 .001 0 009 0 .011 Urba n / Disturbed 0.092 3520.00 0.098 3294.77 0 043 0.098 0.004 0.094 0 .10 2 Wetland 0 026 22.00 0 .001 929 1 9 2.475 0 00 1 0 000 0 000 0 .001 Tota l 1.000 35 790.00 1 000 35790.00 132 734 1 000 Z(l-a/2 K) 2 500 1 = Z(1-a/2K) x (Pi(1 Pi) / n) 112, K = number of habitat categories 2 = Observed u se different from expected when 90 % confidence intervals around observed use falls below (less than expected) or above (more than expected) proportion of habit at w ithin the study area Observed Use Different From Expected2 less less more l ess more less more Jess

PAGE 79

consist of monotypic cultivars rigid stubble and a high percentage of bare ground) Disturbed (unchanged) Weedy forb-Shortgrass combined into shortgrass component Shrubland (unchanged) These five habitat components were further eva l uated based on the pre se nce and absence of prairie dogs. Habitat use-availability analysis of these combined habitats with and without prairie dogs showed that ferruginous hawk did not use habitats in proportion to availability (.x2 = 373.0, df= 9 P < 0.001) (Table 4.2). All habitat components with the exception of shrub l and that had prairie dogs were used more than expected by ferruginous hawks. Conversely the only habitat component without prairie dogs that was used more than expected by ferruginous hawks was the wooded / riparian type. A ll other habitats without prairie dogs were used less than expected. Analysis of Habitat Use From Roadside Counts Goodness of fit comparisons of roadside count data showed that ferruginous hawks preferred (used more than expected) weedy forb areas and avoided ( used less than ex pected) crested w heat g ras s cropland and wetland areas (Table 3.6) Contrar y to results from radio-tracking the roadside counts indicated ferruginous hawks 70

PAGE 80

Table 4 2 Habitat distribution of overwinter i ng ferruginous hawks (n= 17) at Rocky Mountain Arsenal NW A as recorded by radio-trackingtotal observations evaluated with the pre se nce of prairie dogs 1991 94 Proportion Proportion Confidence Interval (90%) Observed Use Habitat of Study Minutes of Total M i nutes Chi-square Lower Upper Different From Type Area Observed Observations (Pi) Expected Contribution Pi zl Limit Limit Expected2 Crested Wheatgrass 0.213 I 085.50 0 030 7620.87 15.658 0.030 0.003 0 028 0.033 less C Wheatgrass w / PO 0.003 392.00 0 011 114.40 1 882 0 .011 0.002 0 009 0 013 more Short Grass 0.485 7602.50 0.212 1 7343.16 15. 284 0 212 0.006 0.206 0.219 less Short Grass w / PO 0.096 16413 00 0.459 3422 23 137. 772 0.459 0 008 0.451 0.466 more Shrub1and 0 058 252.00 0 007 2086 54 4 506 0.007 0 .001 0 006 0 008 less Shrubland w / PO 0.004 71.50 0 002 135.83 0 085 0 002 0.001 0 .001 0.003 less -..l Wooded/Riparian 0 049 5844 50 0.163 1760 65 26.465 0 .163 0 006 0 157 0 169 more _. Wooded/Rip. w / PO 0 000 612.00 0 017 10.71 94.313 0.017 0 002 0 015 0 .019 more Disturbed 0.088 1393.50 0 039 3151.38 2.740 0 039 0 003 0.036 0 042 less Disturbed w / PO 0.004 2126.50 0 059 147 24 74 334 0 059 0 004 0.056 0 063 more Total 1 000 35793.00 1 000 35793 00 373.039 1 000 Z(l-a/2K) 2.576 1 = Z(l-a/2K) x (Pi(1-Pi) / n) 1 1 2 K = number of habitat categories 2 = Observed use different from expected when 90 % confidence intervals around observed use falls below (less than expected) or above (more than expected) proportion of habitat within the s tudy area

PAGE 81

preferred weedy forb areas, although the range of radio-tracking observations was only slightly less than the avai l ability of weedy forb habitats Habitats were then combined into structurally similar habitat components to determine if hawks showed similar habitat selection based primarily on vegetation structure versus composition. Vegetation classes were combined into the five reduced habitat components described above. These habitats were also evaluated based on the presence and absence of prairie dogs Ferruginous hawk use of the combined habitats with and without prairie dogs also did not occur in proportion to availability (Table 4 3). No habitat with prairie dogs present was used less than expected. Three habitats without prairie dogs present (crested wheatgrass shortgrass and s hrub land) were a voided by ferru g inous hawks. Only one habitat shortgrass with prairie dogs was preferred. Observations of ferruginous hawks in all other habitats were in proportion to habitat availability. These results are generally consistent with analysis of radio tracking data. The major difference being that fewer habitats were identified as preferred or avoided using roadside count data. I found that the occurrence of prairie dogs was a major environmental factor influencing habitat selection in ferruginous hawks at RMA. Overall shortgrass habitats were preferred by ferruginous hawks However analysis of habitat use with 72

PAGE 82

-.l Table 4.3 Habitat distribution of overwintering ferruginous hawks on Rocky Mountain Arsenal NWA as recorded by roadside counts observations evaluated with the presence of prairie dogs 1991 94 Proportion Number of Confidence Interval (90 %) Observed use Habitat of Study Number of Observation Chi-square Lower Upper different from Type Area Observation s Expected Contribution Pi z Limit Limit Expected2 C rested Wheatgrass 0.260 14. 50 48 7 2 4 009 0 078 0 050 0 0 27 0 1 2 8 les s C. Wheatgrass w / PO 0.003 3 50 0.6 14.081 0 019 0.026 -0 007 0 044 Short Grass 0.435 56 00 81.3 7 .881 0 299 0.086 0.21 3 0.386 less Short Grass w / PO 0.119 83. 50 22.2 168.780 0.447 0.094 0 .353 0 540 more Shrub land 0 055 3 50 10.3 4 507 0 019 0 026 -0 007 0 044 less Shrubland w / PO 0 002 4.00 0.4 34 774 0 021 0.027 -0.006 0.049 Wooded/Riparian 0.057 7.00 10. 7 1 295 0.037 0 036 0 002 0.073 Wooded/Riparian w / PO 0.001 0.50 0 1 1.545 0.003 0 010 -0.007 0 012 Disturbed 0 067 12.00 12. 5 0 017 0.064 0 046 0.018 0 110 Disturbed w / PO 0 .001 2 50 0.2 32 748 0 013 0 022 -0 008 0 035 Total 1.000 187 00 187 0 289 637 1.000 Z(l-a/2K) 2 576 I = Z(l-a/2K) x (Pi(l-Pi)/ n) 112, K = number of habitat categories 2 Observed use different from expected w h en 90 % confidence intervals around observed use falls below (less than expected) or above (more than expected) proportion of habitat within the study area

PAGE 83

and without the presence of prairie dogs showed that shortgrass with prairie dogs was preferred and shortgrass without prairie dogs was avoided Home Range Results Home range analysis using minimum convex polygons revealed that radio tagged ferruginous hawks occupied relatively distinct home ranges in 1991 / 92 (Figure 4.1 ). Home ranges of ferruginous hawks radio-tracked in 1992 /93 showed much more overlap than 1991192 (Figure 4.2) In the winter of 1993 / 94 radio tracked ferruginous hawks again had relatively distinct home ranges (Figure 4.3). Home ranges varied from 1.51 km2 to 58.13 km2 (Table 4.4). The average size and the standard deviation (SD) of the home range of ferruginous hawks captured at RMA decreased each year ofthe investigation (Table 4.4). Average home range was 29.5 km2 in 1991-92 (n = 6, SD = 23.9 ) 14.4 km2 in 1992-93 (n = 8 SD = 17.4) and 8.3 km2 in 1993-94 (n = 4 SD = 5.2) Wintering ferruginous hawks in 1992 /93 also had relatively smaller home ranges (Figure 4.2). The two individual hawks with the largest home ranges appeared to concentrate activity in a relatively confined area for a period of time then shifted to another activity area. The movements of tagged ferruginous hawks revealed severa l important activity centers at RMA. Two major centers of diurnal activity occurred in the northeastern and southwestern portions of RMA; two areas that supported large 74

PAGE 84

0 0 Figure 4 I Home range of wintering ferruginous hawks r a dio tracked on Rocky Mountain Arsenal NWA in 1991-1992 75

PAGE 85

I 6 Citr i 0 Figure 4 2 Home range of wintering ferruginous hawks radio-tracked on Rocky Mountain Arsenal NWA in 1992 1993 76

PAGE 86

4 I / ..... Cilr I I I I I L _j 0 2 KILOMETERS Figure 4.3 Home range of wintering ferruginous hawks radio -tracked on Rocky Mountain Arsenal NWA in 1993-1994 77

PAGE 87

Table 4.4 Home range calculations using minimum convex poly gons for radio-ta gged ferruginou s hawks at Rock y Mountain Arsena l NW A 1991-94 Bird # Area (km 2) # of Loc a tion s 1991-92 I 58 .13 64 2 1 3 .04 32 3 7.47 110 4 57 .24 39 5 33 95 50 6 7.33 76 Mean 29.53 61.8 3 so 23.90 2 8 55 1992-93 7 9 08 146 8 1.58 17 9 28 .69 198 10 2. 27 27 II 4 7 1 2 0 5 12 50 97 200 13 1.51 33 14 15. 76 54 15 2.3 8 20 Mean 18.24 102.40 so 22.66 92.20 1993-94 1 6 3.4 5 33 1 7 4 .92 60 18 9 .91 41 19 14.86 129 Mean 8 .29 65 .75 so 5 .18 4 3 .66 78

PAGE 88

colonies of prairie dogs. The northwestern portion of RMA was also an important center of activity for a few individual hawks. Description of Communal Roosts Radio-tagged hawks used a total of 29 different roost sites during the three years of study. Sixteen of the 29 roosts were located at RMA. Most of these roosts were used for a short period and by a single radio-tagged hawk ; however several roost sites were used b y numerous hawks for extended per iods in all three years of the investigation. Five important roost sites were identified on RMA based on the frequency of use and the number ofhawks using the roost (Figure 4.4). Measurements were collected on four of these sites. I was unable to access the fifth roost site located adjacent to a communal bald eagle roost due to USFWS restrictions. Wintering ferruginous hawks demonstrated strong roost site fidelity and returned to the same roost 74% of the time (Figure 4.5). Secondary roosts were u sed 21 % of the time and all other roosts were used less then 5 % of the time All ferruginous hawk communal roost sites identified at RMA occurred in isolated groups of trees that permitted access into the roost from all directions and were generally adjacent to open grassland habitats. The number of trees in the roost group ranged from 9 to 44 (Table 4.5). No apparent preference for tree species was 79

PAGE 89

34 35 r I L--' -u' I r-, .. ,. : L-J 51 X T H 561h ..... w j ... a: ....
PAGE 90

5% 0 % Time in main roost % Time in secondary roost % Time in all other roosts -------_j Figure 4 5 Roost site fidelity of ferruginous hawks radio-tracked on Rock y Mountain Arsenal NWR 1991-1994 81

PAGE 91

Table 4.5 Average mea sureme nt s of ferruginous hawk communal roost characteristics Rocky Mountain Arsenal NWA 199194 Roost# DBH Height Height Proximity Distance Dominant (em) ( ft) (m) to Road to P.O. (m) Tree I 23.0 22.8 6.9 17.9 34.6 Elm 2 26.3 26.8 8.2 138. 5 158.5 Elm 3 42 0 34.5 10.5 158. 6 0.0 Cotto nwood 4 123.7 66.8 20.4 160. 2 0.0 Cottonwood 5 ( unable to access roost # 5 due to Bald Eagle Management Area restrictions) 82

PAGE 92

observed as both elms and cottonwoods were used equally. The average tree height per roost ranged from 6.9m to 20.4m with an overall average height of 1 O.Sm. The average DBH per roost was highly variable and was directl y associated with the dominant tree species. Cottonwood domin ate d roo sts were on average taller and wider than elm roost trees. The proximity of roost sites to roads and prairie do g towns was also highly variable at RMA (Table 4 5). Discussion Habitat Use Two data collection techniques (roadside counts and radio-tracking) for the analysis of habitat use and availability both showed that wi ntering ferruginous hawks at RMA avoided crested wheatgrass, cropland and wetland habitats. In addition analysis of radio-tracking data also found that shrub land and weedy forb habitats were avoided. This is contrary to analysis of roadside count data the showed that weedy forb was the onl y habitat preferred b y ferruginous hawks. One limitation of the roadside count is that it only accounts for habitats within a restricted transect width which may or may not contain preferred habitats. In contrast, radio-trackin g has no spatial restrictions and thus can identify all habitats used b y indi vi dual hawks. The contrar y results of th e two techniques could be the effect of ferruginous 83

PAGE 93

hawks selecting for another component of the habitat (i.e. prey) other than landscape composition. Analysis of habitat use and availability evaluated with the presence of prairie dogs showed that wintering ferruginous hawks were strongly associated wi th prairie dogs. Radio-tracking of individual hawks indicated that all habitats containing prairie dogs were preferred and most habitats without prairie dogs were avoided. Roadside counts also showed that shortgrass steppe with prairie do gs was the only habitat preferred by wintering ferruginous hawks. This result is not surprising considering that most prairie dogs occur in the shortgrass steppe. This habitat type comprised about 1 0 percent of the landscape composition of RMA and roadside observation of ferruginous hawks in other habitats containing prairie dogs ma y h ave been insufficient to be statistically significant. Prairie do gs provide a reliable prey source and also modify habitats creating burrows for other prey species such as cottontail rabbits. This reliance or selection of habitat associated with a single prey species has been reported by several researchers Lard y (1980) found that ferruginous hawks selected habitats related to soil types corresponding to the hawks principle prey species the Townsend's ground squirrel (Spermophilus townsendii). Howard and Wolfe ( 197 6) suggested that shifts in vegetation type around successful ferruginous hawk nests were related to jackrabbit densities. 84

PAGE 94

Selection of habitats with prairie dogs may also be a function of prey availability. Black-tailed prairie dogs keep vegetation surrounding their burrows extremely low. This reduced plant cover can increase the detectability and availability of all prey species Previous studies on red-tailed hawks (Preston 1990) and rough-legged hawks (Baker and Brooks 1981) suggest that foraging rap tors respond to the composite of prey density and plant cover density. Wakely (1978) reported that ferruginous hawk foraging distribution was in verse ly relat ed to plant cover density. Wooded / riparian habitat was the only habitat component that was selected b y ferruginous hawks with and without the presence of prairie dog. This is probably a reflection of the preference of wooded areas for roost sites. Other factors that may relate to habitat selection b y wi ntering ferruginous hawks at RMA include foraging behavior detection of prey perch availability and the presence or absence of competitors, particularly bald eagles. Foraging behavior and ultimately habitat selection can be greatly influenced b y perch availability (Janes 1985). In a study of grassland raptors Janes (1985) reported that ferruginous hawks selected areas with relatively few perches and foraged more often from a low soar ing flight while red-tailed hawks selected areas with high perch den sity and foraged more often from perches. 85

PAGE 95

I previously discussed some of advantages and disadvantages of the two data collection techniques used in my study Advantages of roadside counts include ease of establishing and conducting surveys low cost use of basic field equipment and the ability to survey a population rather than a few individuals Disadvantages of roadside counts include spatial and temporal restrictions detectability and the influence of roads on raptor distribution. Radio-tracking has no spatial or temporal restrictions and based on my studies may be able to provide a more accurate and detailed analysis of habitat use. However radio-tracking is labor intensive involves expensive electronic equipment requires large sample size requires more technical skill to operate ma y alter behavior and is stressful to the target species The use of either of these techniques depends on the objectives of the study. Well funded projects that require detailed accurate data and have the opportunity to attach transmitters to an adequate sample size should use radio-tracking techniques. Basic low-budget monitoring studies that need general information or trend data should use standard roadside counts Home Range The reduction in size of ferruginous hawk home ranges corresponds with the increase in average number of wintering ferruginous hawks observed p e r road transect. Both these observations coincide with the increase in prairie dogs at RMA (D. Serri pers. comrnun.). This suggests th a t ferruginous haw ks decrea s ed the 86

PAGE 96

average distance traveled in each year of the study, probably in response to the increased prey availability at RMA and thus an increase in the area of the preferred habitat type (shortgrass steppe with prairie dogs). By decreasing the distance traveled to obtain necessary resources (food and cover) wintering ferruginous hawks could reduce energy expenditures and increase survivability during this critical period. Increasing density of ferruginous hawks in response to increased prey confirms studies conducted by Schmutz and Hungle (1989) who found that the density of breeding ferruginous hawks increase with ground squirrel abundance This indicates that when hawks are able to satisfy food requirements within a smaller area a higher density is tolerated. Thus, the increase in ferruginous hawks observed per road count transect on RMA could be due either to an increase in the number and density of hawks using RMA, or from ferruginous hawks constricting their home range and spending more time at RMA, in both cases a response to increased prey density. Home range size estimation is subject to a researcher's choice of home range model and is dependent on the method of data collection. Observations collected in a short sampling period often are autocorre l ated (Swihart and Slade 1985) which can seriously bias results. The objectives of my study were to determine home-range boundaries, specifically home-range of ferruginous hawks wintering at RMA. Short sampling intervals are essential in this type of study and non-statistical estimates 87

PAGE 97

(polygon measures) are recommended to minimize the effects of autocorrelation (Swihart and Slade 1985). I found that ferruginous hawk home ranges typically extended be y ond the boundaries ofRMA. Frequently hawks roosted at RMA foraged throughout the day off-site then returned to roost at RMA. Thus RMA may be regionally important to wintering ferruginous hawks and conversely, off-site foraging areas particularly Stapleton International Airport and open fields in south Adams County appear to be essential or important resources to the management of ferruginous hawks at RMA. This is especially pertinent in view of recent developments including sylvatic plague that has decimated RMA prairie dog populations the closing of Stapleton International Airport and urban growth in Adams County that has eliminated man y of the foraging and off-site roosting areas described in this report. Communal Roost Communal roosts have been widely studied in bald eag les (Keister and Anthony 1983 Grubb et al. 1989 Buehler et al 1991) and common ravens (Engel et al. 1992) but has received little attention for ferruginous hawks (Steenhof 1984 ) In fact few studies have been conducted on wintering ferru g inous hawks in g eneral which may account for the paucity of liter ature on communal roosting behavior. 88

PAGE 98

Two aspects of ferruginous hawk communal roosting behavior were observed at RMA that are similar to behaviors of other communally roosting species (Eiserer 1984). First the same communal roosts were used year after y ear. Main communal roosts specifically in sections 5 9 11, and 25, were used in successive y ears b y ferruginous hawks wintering at RMA. At least three of these roosts were consistently used b y ferruginous hawks prior to the stud y ( Beane pers. obs. ) Second the number of birds at communal roosts fluctuated seasonally and yearl y (Eiserer 1984). The number of ferruginous hawks in a particular RMA roost varied throughout the study and observations of radio-tagged hawks indicated that some individual hawks moved between roosts. Although some movement between roosts was observed overall roost site fidelity b y ferruginous hawk s wintering at RMA was strong. I frequently observed hawks flying past an established communal roost to reside for the night in a more distant primar y roost. Although the adaptive significance of communal roosting is not well understood several hypothesis have been developed to explain this behavior. These hypothesis include selection of microclimates that provide energetic benefits (Keister et al 1985 ), and association with concentrated or sporadicall y abundant food resources (Engel et al 1992 Steenhof 1976 ) In addition to microclimate communal roosts may conserve energy as the raptors roost in the nearest roost site to prey resources. Raptors such as bald eagles that are adapted to exploit seasonall y 8 9

PAGE 99

concentrated food resources (e .g., salmon) or depend on a single prey species ma y be more likely to concentrate in communal roosts at seasonally or cyclically abundant food resources. The adaptive significance of this behavior may include use of the roost as information centers that alert roost members of available prey or allow young to gain from the experience of adults in the roost (Ward and Zahavi 1972). The ferruginous hawk in eastern Colorado is strongly associated with prairie dogs a species that historically and currently experience wide population fluctuations and cyclic plague outbreaks Considering this fluctuating pre y base it would be adaptive for wintering ferruginous hawks to 'follow the crowd and roost near abundant prey resources 90

PAGE 100

CHAPTER 5 OVERALL DISCUSSION AND CONCLUSIONS Documenting relative abundance and distribution of raptors can detect trends or determine general patterns of habitat use but provides little insight into the specific habitat components selected by diurnal raptors Further investigation into habitat use and availability based on landscape composition provides more information on habitat selection. However important components within habitats may still be overlooked. Principal component analysis can help the researcher ordinate species with respect to specific components, or habitat gestalt within the more general habitat classifications My studies used a series of relative abundance and distribution studies combined with habitat use and availability studies and PCA to provide a clear picture of diurnal raptor habitat use at RMA. These studies helped me achieve my five primary objectives and reach the following conclusions: 1. raptors at RMA clearly indicate that this area is regionally important for raptors. 2. The distribution abundance and habitat use of overwintering ferruginous The distribution and abundance of both breeding and wintering diurnal hawks is dependent on a combination of integrated habitat components consisting of; 1) 91

PAGE 101

abundant and available prey (prairie dogs); 2) low-growing grassland or weedy forb habitat; 3) low vegetation cover and diversity; 4) few perches ; and 5) available roost sites in proximity to prey. 3. Prey abundance and availability is the major habitat component that determines distribution, abundance, habitat use home range size, and roost site selection by overwintering ferruginous hawks. 4 Overwintering ferruginous hawks tend to use communal roosts, and frequently return to the same roost throughout the winter. 5. Crested wheatgrass and cropland habitats are avoided by most diurnal raptors in both summer and winter. 6. Both summer and winter diurnal raptor communities do not use habitat in proportion to availability and segregate along distinct habitat gestalts. The habitat gestalts for summer diurnal raptors at RMA consist of: American kestrel diverse weedy vege tation with few perches Red-tailed hawk-wooded areas with numerous perches Swainson s hawk-more monotypic vegetation with few perches or wooded areas 92

PAGE 102

The habitat ges talt for winter diurnal raptor s at RMA consist of: Ferruginous hawk-areas with low vegetation cover and diversity few trees or other perches abundant prairie dogs and nearby roost sites. Red-tailed hawk-woodland/wetland habitats with numerous perches and relatively low vegetation cover and diversity. Rough-legged hawks-wetland habitat with fewer trees and perches than habitats used by red tailed hawks or bald eagles but more than habitats used by ferruginous hawks. Bald eagle woodland habitat with numerous large trees and perche s, typically associated with water features and high vegetation cover and diversity 93

PAGE 103

REFERENCES CITE D Andersen D.E. and 0. J. Rongstad. 1989. Home range estimates of red-tailed hawks based on random and systematic relocations. J. Wild!. Manage. 53:802-807. Andersen D.E ., 0.1. Rongstad and W.R. Mytton. 1986. The behavioral response of a red-tailed hawk to military training activity. Raptor Res. 20:65-68 Andersen D.E. 0.1 Rongstad and W.R. Mytton 1989. Response of nesting red tailed hawks to helicopter overflights. Condor 91:296-299. Andersen D.E., O.J. Rongstad and W.R. Mytton. 1990 Home range changes in raptors exposed to increased human activity in southwestern Colorado Wild!. Soc. Bull. 18:134-142 Baker J.A. and R.J Brooks. 1981 Distribution patterns of raptors in relation to density of voles. Condor 83:42-47 Bechard M J. 1982. Effect of vegetative cover on foraging site selection by Swainson's hawk. Condor 84 : 153-159 Bechard M.J., R.L. Knight, D.G Smith and R.E. Fitzner. 1990. Nest sites of sympatric hawks (Buteo spp.) in Washington. J. Field Ornith. 61:159170 Bechard M.J. and J.K. Schmutz. 1995. Ferruginous hawk (Bu t eo regalis) in A. Poole and F. Gill [ E ds ] The Birds ofNorth America, No. 172. The Academy of atural Sciences, Philadelphia, and The American Ornithologists Un ion Washington D C. Bitstream Inc. 1991. Harvard Graphics Software Publishing Corporation. Bloom P. H. 1987. Capturing and handling techniques Pages 99-123 in B.A. Giron B .A. Millsap K.W Kline and D.M Bird [Eds.] Raptor management techniques manual. Natl. Wildl. Fed ., Sci. and Tech Ser. No. 10. 94

PAGE 104

Bockus D.B. 1994 Diet of wintering ferruginous hawk s at Rocky Mountain Arsenal. Unpublished MSES project draft report. October 5 Boone J.D. and C.R. Preston. 1994. Documentation and interpretation of selected wildlife habitat relationships at the Rocky Mountain Arsenal: t ask one small rodents. Prepared by the Denver Museum ofNatural Hist o r y for the U.S. Fi s h and Wild!. Serv March. Buehler D.A. T.J Mersmann J.D Fraser and J.K.D. Seegar. 1991. Nonbreeding bald eagle communal and so lit ary roosting behavior and r o ost habitat on the northern Chesapeake Bay 1. Wild!. Manag e 55: 273 -2 81. Byers C. R. and R.K Steinhorst. 1984. Clarification of a technique for anal ys i s of utilization-availability data. 1. Wildl. Manage. 48 : 1050-1053 Cottrell M.J. 1981. Resource partitioning and reproductive success of hawks (Buteo spp ) in an Oregon prairie. M.S thesis. Oregon State Univ., Corvallis. Craighead J.J. and F.C. Craighead. 1956. Hawks owls and wildlife. Dover Publications Inc., New York NY. Craighead J.J. 1987. Forward in B.A. Giron Pendleton B.A. Millsap K W Cline, and D. M. Bird [Eds.] Raptor management techniq u es manual. Natl. Wildl. Fed., Washington D.C. Dixon K R., and J.A. Chapman 1980 Harmonic Mean measure of animal activity areas Ecology 61:1040-1044. Eiserer L.A. 1984. Communal roosting in birds. Bird Behav. 5:61-80. Engel K.A. L.S. Young K. Steenhof J.A. Roppe, and M.N Kochert. 1992. Communal roosting of common ravens in southwestern Idaho. Wilson Bull. 104 : 105-1 21. Environmental Science and Engineering (ESE) 1989 Biota remedial investigation final report. Prepared for Program Manager RMA Contamination Cleanup. Rocky Mountain Arsenal. Commerce City CO 95

PAGE 105

Fuller, M.R. and J.A. Mosher. 1981. Methods of detecting and counting raptors: a review. Pages 235-246 in C.J. Ralph and J.M Scott [Eds.] Esti matin g numbers of terrestrial birds. Stud. A vi an Bioi. 6 Fuller M.R. and J.A. Mosher. 1987. Raptor survey techniques. Pages 81-98 in B.A. Giron Pendleton B.A. Millsap K .W. Kline and D M Bird [Eds] Raptor management techniques manu al. Vol. I. Nat!. Wild!. Fed. Sci. and Tech. Ser. No. 10. Grubb T.G. and R.M King. 1991. Assessing human disturbance of breeding bald eagles with classification tree models J. Wild!. Manage. 55 : 500511. Holmes T.L., R.L. Knight, L. Stegall and G R Craig 1993 Responses of wintering grassland raptors to human disturbance. Wild!. Soc Bull. 21 :461-468. Howard R.P. and M .L. Wolfe. 1976 Range improvement practices and ferruginous ha wks. J. Ran ge Manage. 29 :33-37. Houston, C.S and M.J. Bechard. 1984. Decline ofthe ferruginous hawk in Saskatchewan. Am. Birds 38:166-170. Hutto R. L. 1985 Habitat selection by non-breeding migratory land birds. Pages 455-476 in M.L. Cody [Ed ] Habitat selection in birds Academic Press Inc., San Diego CA. James F. C. 1971. Ordinations of habitat relationships among breeding birds. Wilson Bull. 83:215-236. Janes, S.W. 1985. Habitat selection in raptorial birds, Pages 159-188. in M.L. Cody [Ed.], Habitat selection in birds. Academic Pres s Inc. San Diego CA. Keister, G.P and R.G. Anthony. 1983. Characteristics ofbald eagle communal roosts in the Klamath Basin Oregon and California. J. Wildl. Manage. 47:1072-1079. 96

PAGE 106

Keister G.P., R.G Anthony H.R Holbo. 1985. A model of energy consumption in bald eagles: an evaluation of night communal roost ing. Wilson Bull. 97:148-160 Kenward R.E. 1987 Wildloife radio tagging. Acedemic Press Inc ., San Die g o CA. Kimsey B. and M.R. Conle y 1989. Habitat use by raptors in southcentral N e w Mexico. Pages 197-203 in R.L. Glinski B.G Pendleton M B. Moss M.N. LeFranc Jr. B.A. Milsap and S.W Hoffman [Eds.] Proc southwest raptor management s y mpos ium and workshop. Natl. Wildl. F ed. Washin g ton D.C. Kingery H 1996 Unpublished results of the Denver Urban Chr i stmas Bird C ount ; 1989 1995 J a nuary 1. Knight R.L. and S.K. Knight. 1984. Responses of wintering bald eagles to boating activity. J. Wildl. Manage 47:999-1004 Knight R L. and S.K Skag en. 1988. Effect s o f recreational disturbance o n bird s of prey : a review Page s 355-359 in R .L. Glinski B G Pendleton M.B Moss M.N. LeFranc Jr. B.A. Milsap and S .W. Hoffman [Eds.] Proc. southwestern raptor management symposium and workshop. Natl. Wildl. Fed., Washington DC. Kochert M N. 1986. Raptors Pages 313-350 in A.Y Cooperrider, R.J. Boyd and H.R Stuart [Eds.] Inventory and monitoring of wildlife habitat. U.S Dep Inter. Bur. Land Management. Service Center. Denver CO Lardy M E 1980 Raptor inventory and ferruginous hawk breeding biology in southeast Oregon. M S thesis. University ofldaho, Moscow. Lauenroth W.K. and D .G. Milchunas. 1991. Short-grass steppe. Chapter 11 in R T. Coupland [Ed ], Ecosystems ofthe world; natural grass lands. Elsevier. Amsterdam Millsap B.A. and M.N. LeFranc Jr. 1988. Road transect counts for raptor s : how reliable are they ? J. Rapt. Res. 22:8-16. 97

PAGE 107

Mohr C.O. 1947. Table of equivalent populations of North American s mall mammals. Am. Midi. Nat. 37:223-249 Morrison-Knudsen Environmental, Inc. (MKE) 1989. Wildlife resources of the Rocky Mountain Arsenal Adams County Colorado. Prepared for Shell Oil Company Denver Co August. Neu C.W., C .R Byers and J.M. Peak. 1974. A technique for analysis of utilization availability data. J. Wildl. Manage. 38 : 541-545. Newton I. 1979 P opu lation ecology ofraptors. Bute o Books. Vermillion SD Olsen S.J. 1964. Mammal remains from archeological sites : Part 1 : Southeastern and southwestern United States. The Peabody Museum, Cambridge, MA. Olendorff, R.R. 1972. The large birds of prey of the Pawnee National Grassland: Nesting habits and Productivity 1969-1971. IBP Grassland Biome Tech. Rep. 151. Natural Resources Eco l ogy Laboratory. Colorado State University Fort Collins CO. Olendorff R.R. 1973. Ecology of the nestin g birds of pre y of northeastern Colorado. IBP Grassland Biome Tech. Rep 211, Natural Resources Ecology Laboratory. Colorado State U niversity Fort Collins, CO Olendorff R .R. 1993. Status biology and management offerruginous hawk: a review. Raptor Res and Tech Asst. Cen., Spec. Rep U.S. Dep. Interior, Bur Land Manage. Boise ID. Partridge L. 1978 Habitat selection Pages 351-376 in J.R. Krebs and N.B. Davies [Eds ] Behavioral ecology-an evo lution ary approach. Sinauer Press Sutherland MA. Per Widen P. 1994. Habitat quality for raptors : a field experiment. J. Avian Bioi. 25:219-223. Pianka E.R. 1978. Evolutionary ecology. Harper and Row New York NY. 98

PAGE 108

Preston C.R. 1990. Distribution of raptor foraging in relation to prey biomass and habitat s tructure Condor 92: I 07-112. Pre ston, C.R., and R.D. Beane 1996. Occurrence and distribution of diurnal raptors in relation to human activity and other factors at Rocky Mountain Arsenal Colorado. Pages 365-374 in D .M. Bird D Varland J.J Negro [Eds.] Raptors in Human Landscapes. Academic Press Ltd. San Die go, CA. Restani M. 1991. Resource partitioning among three buteo species in the Centennial Valley Montana Condor 93:1007-1010. SAS Institute Inc Cary, C. ( Ver s ion 5) Schmutz J.K 1984. Ferruginous and Swainson's hawk abundance and distribution in relation to land use in so uthea stern Alberta. J Wildl. Manage. 48 : 1180-1187 Schmutz, J.K., S.M. Schmutz and D .A. Boag. 1980 Coex istence ofthree species of hawks (Buteo spp.) in the prairie-parkland ecotone. Can. J. Zool. 58:1075-1089. Schmutz J.K. and D .J. Hungle. 1989. Population of ferruginous and Swainson's hawks increase in synchrony with ground squirrels Can. J Zool. 67:2596-2601. Schnell G.D. 1968. Differential habitat utilization b y wintering rough-leg ged and redtailed hawks. Condor 70:373-377. Sherrod S.K. 1978. Diets of orth American Falconiformes. Rapt. Res 12:49121. Smith D.G., and J.R Murphy. 1973 Breeding ecology ofraptors in Utah. Brigham Young Univ. Sci Bull. Bioi. Ser. 18(3):1-76. Southwood, T.R.E. 1966. Ecological methods with particular reference to th e study of insect populations. Meuthuen and Co. Ltd London, U.K. Stalmaster M.V. and J .R. Newman. 1978. Behavioral responses of wintering bald eagles to human activity J. Wildl. Manage. 42:506-513. 99

PAGE 109

Steenhof K. 1984 Use of an interspecific communal roost by wintering ferruginous hawks. Wilson Bull. 96:137-138. Steenhof, K. 1976. The ecology of wintering bald eagles in so uthea stern South Dakota. M.S. thesis, univ Missouri Co lumbia 148 pp. Stuwe, M. 1992. McPaal Homerange Software Version 1.22 Swihart R.K. and N A. Slade 1985. Influence of sampling interval on estimates of home-range size. J. Wild!. Manage. 49:1019-1025. Terborgh, J. 1992. Perspectives on the conservation of neotropical migrant landbirds. Pages 7-12 in J.M. Hagan and D.W. Johnson [Eds.], Eco log y and conservation of neotropical migrant land birds. Smithsonian Institution Press Washington D .C. Tinbergen, N. 1951. The study of instinct. Oxford University Press London. U.S. Fish and Wildlife Service. (USFWS). 1994 U.S. Fish and Wildlife Service Rocky Mountain Arsenal National Wildlife Refuge fiscal year 1993 annual progress report. Building 111, Rock y Mountain Arsenal NWR, Commerce City CO U.S. Fish and Wildlife Service. (USFWS) 1995. U.S. Fish and W ildl ife Service Rocky Mountain Arsenal National Wildlife Refuge fiscal year 1994 annual progress report. Building 111, Rock y Mountain Arsenal NWR, Commerce City, CO. February 15. U.S Fish and Wildlife Service (USFWS). 1996. Rocky Mountain Arsenal National Wildlife Refuge final environmenta l impact statement. Rocky Mountain Arsenal NWR. Building 613 Co mmerce City CO. January. Wakeley U S 1978 Factors affecting the use of hunting sites b y ferrugino u s hawks Condor 80:316-326 Ward, P. and A. Zahavi 1 972. The importance of certain assemb l ages of birds as "information centers" for food finding. Ibis 115 :51 7 5 34. 100

PAGE 110

White G.C. and R.A. Garrott. 1990. Analysis of radio-tracking data. Academic Press Inc. San Diego, CA 101