THE CONSERVATION RESERVE ENHANCEMENT PROGRAM (CREP) AND
GRASSLAND BIRD CONSERVATION IN MICHIGAN

By

Adria Stilwell VanLoan

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Fisheries and Wildlife and Ecology, Evolutionary Biology, and Behavior

2011
i

ABSTRACT
THE CONSERVATION RESERVE ENHANCEMENT PROGRAM (CREP) AND
GRASSLAND BIRD CONSERVATION IN MICHIGAN
By
Adria Stilwell VanLoan
Grassland birds are one of the most imperiled guilds of North American birds, and
their populations continue to decline in Michigan. Habitat loss and fragmentation due to
agricultural conversion have been the most important causes of decline, with losses of
native temperate grasslands >83% in the Midwest. Farmland set-aside programs such as
the Conservation Reserve Enhancement Program (CREP) increase the area of native and
introduced grasslands in the U.S., contributing to the conservation of grassland birds.
Local evaluations (e.g., within major watersheds or ecoregions) of the impacts of farmland
set-aside programs on grassland bird populations are important because grassland bird
habitat requirements vary among geographic regions and there is geographic variation in
the effect of farmland set-aside lands on wildlife. I examined avian populations of CREP
grasslands in the Saginaw Bay watershed of Michigan from 2005-06 to address the
following questions: 1) Are grassland birds present? 2) Do densities of grassland species
differ in native and introduced CREP grasslands? 3) Is grassland bird occupancy correlated
with grassland size, type, and structural variables and/or the extent or proximity of woody
vegetation near grasslands (≤100 m)? Results of this study show that CREP grasslands
provide habitat for grassland bird species in Michigan. Of 15 grassland species
documented to occur in the state in the most recent Michigan Breeding Bird Atlas, 11
species were observed in study grasslands, including 1 Michigan endangered species
ii

(Henslow‘s Sparrow), 3 Michigan special concern species (Dickcissel, Grasshopper
Sparrow, and Northern Harrier), and 4 species with significantly declining population
trends (p < 0.10) in Michigan according to recent analyses of Breeding Bird Survey data
(Bobolink, Eastern Meadowlark, Ring-necked Pheasant, and Vesper Sparrow). Detection
probability estimates were calculated for observed species. Species with low detection
probability estimates (i.e., < 0.10) were not included in comparisons of density or in
occupancy modeling. Densities of 5 of 8 grassland bird species with detection probability
estimates ≥ 0.10 were higher in native grasslands than introduced grasslands during at least
one year of the study (i.e., Grasshopper Sparrow, Ring-necked Pheasant, Savannah
Sparrow, Sedge Wren, and Vesper Sparrow). One species, Bobolink, was more abundant
in the introduced vegetation during one year of the study. Grassland type was positively
associated with the occupancy of Ring-necked Pheasant, Sedge Wren, and Vesper Sparrow,
with higher detection-corrected occupancy estimates in native grasslands for all species.
Grassland size was positively associated with Sedge Wren occupancy and negatively
associated with Dickcissel occupancy. Although grassland type and grassland size were
not significant correlates of occupancy for any species, results suggest that these covariates
are more than random effects of grassland bird occupancy and hence should be included in
future studies and considered in grassland bird conservation strategies. Occupancy
modeling results also suggest a diverse response of grassland bird species to habitat
features, indicating that species-specific information should be used when managing
grassland bird species. Study results demonstrate that CREP grasslands, particularly native
grasslands, provide grassland bird habitat, suggesting that maintaining and increasing the
area of CREP grasslands should be a grassland bird conservation strategy in Michigan.

iii

This work is dedicated those who protect and care for little brown birds and other
grassland species.

iv

ACKNOWLEDGMENTS

This research was funded by the Michigan Department of Natural Resources,
Wildlife Division. Other financial support was provided to me by the Michigan State
University Graduate School, through the University Distinguished Fellowship, by the
MSU Program in Ecology, Evolutionary Biology, and Behavior, through the EEBB
Summer Fellowship, and by the MSU Department of Fisheries and Wildlife, through a
teaching salary. The Michigan Department of Agriculture provided Conservation
Reserve Enhancement Program information that was critical to this research, and Tuscola
County Soil Conservation District provided landowner introductions and technical
information. My thanks go to all of these entities.
Many people supported the completion of my dissertation research, and I am
grateful to each one of them. My major advisor, Kelly Millenbah, and my other
dissertation committee members, Rique Campa, Catherine Lindell, and Gary Roloff,
along with Kay Gross and Brian Maurer, provided me with valuable knowledge, advice,
and encouragement. Bill Taylor, former Chair of the Fisheries and Wildlife Department,
helped launch me at MSU and Mike Jones, current Fisheries and Wildlife Department
Chair, helped me navigate and complete my journey. Other MSU faculty and staff also
generously assisted and encouraged me, including Marcia Baar, Jim Brown, Jill Cruth,
Dan Hayes, Josh Hulst, Dana Infante, Susan Jones, Doug Landis, Fred Poston, Sharon
Reasoner, Julie Traver, and Mary Witchell. My fellow Fisheries and Wildlife
Department graduate students and labmates, Lauren Bailey, Tracy Grazia, Katie Kahl,

v

Nikki Lamp, Sarah Panken, Nichole Rubeck-Schurtz, and Trixi Smith, provided me with
advice, moral support, and camaraderie.
Mike Donovan and Mark Sargent from the Michigan Department of Natural
Resources and Chris Savona and Steve Shine from the Michigan Department of
Agriculture were especially helpful in the planning of my dissertation research. Jim
Kratz and the other generous souls at the Tuscola County Soil Conservation District
provided assistance in the execution of this research in a variety of ways, from helping to
identify potential study sites and secure landowner permission to sharing plant
identification and other technical knowledge. The hard work and diligence of field
technicians Jessie Avery, Megan Bowman, Phyllis McKenzie, and Chris Vander Heyden,
as well as frequent field crew volunteer and exceptional mom Mary VanLoan, were
essential to this project.
Completion of this research and the other requirements of my Ph.D. program
would have been impossible without the loving support of my family and friends, whom I
deeply appreciate. To Lauren Bailey, Laurie Brooks, Mrs. Burroughs, Pete and Denise
Castor, Tanya Denyes, Anna Fiedler, Freedom Raymond, Terry Rosander, Lee Anne
Sims, and Eric Trevan, thank you for your friendship and encouragement. To my
parents, Chuck and Mary, my brother and sister-in-law, Mike and Jenny, my sister,
Elizabeth, my grandmothers, Gladys and Lillian, my aunts, uncles, and cousins, my
father-in-law, Mark, Honey and my other animal companions, and last but not least, my
husband, Luke, who shared this journey with me more closely than anyone, thank you for
everything.

vi

TABLE OF CONTENTS

LIST OF TABLES.......................................................................................................

ix

LIST OF FIGURES......................................................................................................

xiv

CHAPTER 1: INTRODUCTION, RESEARCH OBJECTIVES, AND STUDY
AREA
Introduction......................................................................................................
Research Objectives.........................................................................................
Study Area........................................................................................................
Appendix 1.1: Chapter 1 Tables and Figures...................................................
Literature Cited.................................................................................................

1
4
5
8
11

CHAPTER 2: GRASSLAND BIRD DETECTION PROBABILITY ESTIMATION
AND DENSITY COMPARISON IN MICHIGAN CONSERVATION RESERVE
ENHANCEMENT PROGRAM (CREP) GRASSLANDS
Introduction......................................................................................................
Methods............................................................................................................
Study Area............................................................................................
Vegetation Sampling............................................................................
Vegetation Variable Comparison.........................................................
Avian Surveying..................................................................................
Avian Density Comparison..................................................................
Detection Probability Modeling...........................................................
Results..............................................................................................................
Vegetation............................................................................................
Grassland Birds....................................................................................
Detection Probability Modeling...............................................
Density Comparison.................................................................
Discussion.........................................................................................................
Vegetation.............................................................................................
Grassland Birds....................................................................................
Management Recommendations..........................................................
Appendix 2.1: Chapter 2 Tables and Figures...................................................
Appendix 2.2: Detection Probability Modeling and Density Comparison
Results for Avian Species Other than Grassland Specialists................
Appendix 2.3: Comparison of Avian Productivity within CP23 Native and
CP1 Introduced Plant Communities of Michigan Conservation
Reserve Enhancement Program (CREP) Grasslands...........................
Appendix 2.4: Comparison of Avian Species Presence in Michigan
Conservation Reserve Enhancement Program (CREP) and
Conservation Reserve Program (CRP) Introduced Grassland
Vegetation.............................................................................................
Literature Cited.................................................................................................
vii

17
19
19
20
21
22
22
23
25
25
30
31
33
38
38
39
42
45
78

91

97
105

CHAPTER 3: GRASSLAND BIRD DETECTION AND OCCUPANCY IN
MICHIGAN CONSERVATION RESERVE ENHANCEMENT PROGRAM
(CREP) GRASSLANDS
Introduction......................................................................................................
Methods............................................................................................................
Study Area............................................................................................
Avian Surveying...................................................................................
Detection Probability and Occupancy Modeling.................................
Results..............................................................................................................
Detection Probability Modeling...........................................................
Occupancy Modeling...........................................................................
Discussion.........................................................................................................
Detection Probability and Occupancy Modeling.................................
Management Recommendations..........................................................
Appendix 3.1: Chapter 3 Tables and Figures...................................................
Appendix 3.2: All Avian Species Observed.....................................................
Appendix 3.3: Occupancy Modeling Results for Avian Species Other Than
Grassland Specialists............................................................................
Literature Cited.................................................................................................
CHAPTER 4: GRASSLAND BIRD DETECTION AND OCCUPANCY IN
NATIVE AND INTRODUCED PLANT COMMUNITIES OF MICHIGAN
CONSERVATION RESERVE ENHANCEMENT PROGRAM (CREP)
GRASSLANDS
Introduction......................................................................................................
Methods............................................................................................................
Study Area............................................................................................
Avian Surveying...................................................................................
Detection Probability and Occupancy Modeling.................................
Results..............................................................................................................
Detection Probability Modeling...........................................................
Occupancy Modeling...........................................................................
Discussion.........................................................................................................
Detection Probability and Occupancy Modeling.................................
Management Recommendations..........................................................
Appendix 4.1: Chapter 4 Tables and Figures...................................................
Appendix 4.2: All Avian Species Observed.....................................................
Literature Cited.................................................................................................

113
114
114
116
117
119
120
121
124
124
127
130
144
148
157

162
163
163
165
165
171
172
173
176
176
181
184
203
207

CHAPTER 5: SUMMARY AND MANAGEMENT RECOMMENDATIONS
Summary and Management Recommendations............................................... 214
Literature Cited................................................................................................. 223

viii

LIST OF TABLES

Table 1.1. Plant species and proportions of species included in the seed mixtures
used to establish the native and introduced vegetation communities of Conservation
Reserve Enhancement Program (CREP) grasslands of Michigan‘s Saginaw Bay
watershed, 2002...........................................................................................................

10

Table 2.1. Plant species and proportions of species included in the seed mixtures
used to establish the native and introduced vegetation communities of Conservation
Reserve Enhancement Program (CREP) grasslands of Michigan‘s Saginaw Bay
watershed, 2002...........................................................................................................

47

Table 2.2. Plant species observed in native and introduced plant communities of
Conservation Reserve Enhancement Program (CREP) grasslands, Saginaw Bay
watershed, Michigan, 2005 and 2006..........................................................................

48

Table 2.3. Avian species observed in CP23 native fields, the introduced portion of
CP1 fields, and CP1 whole fields, Conservation Reserve Enhancement Program
(CREP) grasslands, Saginaw Bay watershed, Michigan, 2005-2006. Avian
densities (# individuals /ha [SE]) are given for species with detection probability
estimates ≥0.10. Species observed too rarely to support detection probability
modeling (i.e., observed in < 10% of surveys) or with detection probability
estimates <0.10 are noted as present when observed in at least one survey and as
undetected when unobserved.......................................................................................

58

Table 2.4. Grassland bird species, with corresponding best-fitting detection models
[with constant occupancy, ψ(·)] and detection probability estimates, observed in 1)
CP23 and CP1 Conservation Reserve Enhancement Program (CREP) grassland
fields and 2) CP23 native and CP1 introduced grassland plant communities,
Saginaw Bay watershed, Michigan, 2005-2006. The lowest detection probability
estimate was reported when multiple detection probability estimates were
generated......................................................................................................................

61

Table 2.5. Grassland bird species, with estimated Michigan breeding territory size
range (ha) and estimated number of breeding territories in CP23 and CP1
Conservation Reserve Enhancement Program (CREP) fields and CP1 introduced
grassland plant communities, Saginaw Bay watershed, Michigan, 2005-2006...........

77

Table 2.6. Non-grassland avian species, with corresponding best-fitting detection
models [with constant occupancy, ψ(·)] and detection probability estimates,
observed in 1) CP23 and CP1 Conservation Reserve Enhancement Program
(CREP) grassland fields and 2) CP23 native and CP1 introduced grassland plant
communities, Saginaw Bay watershed, Michigan, 2005-2006. The lowest detection
probability estimate was reported when multiple detection probability estimates
were generated.............................................................................................................

84

ix

Table 2.7. Active avian nests observed, nest outcome, and vegetation type of the
nest location in Conservation Reserve Enhancement Program (CREP) whole field
grasslands, Saginaw Bay watershed, Michigan, 2006.................................................

96

Table 2.8. Mean [SE] vegetative characteristics of 1.) introduced Conservation
Reserve Enhancement Program (CREP) grasslands, Saginaw Bay watershed,
Michigan, mid-summer (15 June – 14 July) 2005 and mid-summer 2006 (n = 7),
and 2.) Conservation Reserve Program (CRP) grasslands, Gratiot County,
Michigan, July 1991 and July 1992 (n = 3). CREP fields were planted in 2002 and
CRP fields were planted in 1988..................................................................................

100

Table 2.9. Avian species observed in 1.) introduced vegetation of Conservation
Reserve Enhancement Program (CREP) grasslands, Saginaw Bay watershed,
Michigan, 2005-2006, and 2.) introduced vegetation of Conservation Reserve
Program (CRP) grasslands, Gratiot County, Michigan, 1991-1992. Avian densities
(# individuals/ha) [SE] are given for species with successful detection probability
modeling (p≥0.10). In CREP vegetation, species observed too rarely to support
detection probability modeling (i.e., observed in < 10% of surveys) or with
detection probability estimates <0.10 are noted as present when observed in at least
one survey and as undetected when unobserved. Detection probability modeling
was not performed for avian species observed in CRP vegetation..............................

101

Table 3.1. Grassland bird species, with corresponding best-fitting detection models
(with constant occupancy, ψ[·]) and detection probability estimates, observed in
CP23 and CP1 Conservation Reserve Enhancement Program (CREP) grasslands,
Saginaw Bay watershed, Michigan, 2005-2006. The lowest detection probability
estimate was reported when multiple detection probability estimates were
generated......................................................................................................................

132

Table 3.2. Grassland bird species and global occupancy models with
corresponding number of parameters, K, and overdispersion estimate, ĉ, observed
in CP23 and CP1 Conservation Reserve Enhancement Program (CREP) grasslands,
Saginaw Bay watershed, Michigan, 2005-2006..........................................................

138

Table 3.3. Summary of AIC occupancy model selection adjusted by the best
detection model for Dickcissel in CP23 and CP1 Conservation Reserve
Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan,
2005 - 2006..................................................................................................................

139

Table 3.4. Parameter estimates for the best fitting occupancy models for Dickcissel
and Grasshopper Sparrow and model averaged parameter estimates for Sedge Wren
(2006 data only) and Vesper Sparrow in CP23 and CP1 Conservation Reserve
Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan,
2005-2006....................................................................................................................

140

x

Table 3.5. Summary of QAIC occupancy model selection adjusted by the best
detection model for Grasshopper Sparrow in CP23 and CP1 Conservation Reserve
Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan,
2005-2006....................................................................................................................

141

Table 3.6. Summary of AIC occupancy model selection adjusted by the best
detection model for Sedge Wren, 2006 data only, in CP23 and CP1 Conservation
Reserve Enhancement Program (CREP) grasslands, Saginaw Bay watershed,
Michigan......................................................................................................................

142

Table 3.7. Summary of AIC occupancy model selection adjusted by the best
detection model for Vesper Sparrow in CP23 and CP1 Conservation Reserve
Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan,
2005-2006....................................................................................................................

143

Table 3.8. Non-grassland avian species, with corresponding best-fitting detection
models (with constant occupancy, ψ[·]) and detection probability estimates,
observed in 1) CP23 and CP1 Conservation Reserve Enhancement Program
(CREP) grasslands and in 2) CP23 fields and the introduced portion of CP1 fields,
Saginaw Bay watershed, Michigan, 2005-2006. Ordinal date was standardized to
reduce biases in parameter estimation. The lowest detection probability estimate
was reported when multiple detection probability estimates were generated.............

151

Table 3.9. Non-grassland avian species and global occupancy models with
corresponding number of parameters, K, and overdispersion parameters, ĉ,
observed in CP23 and CP1 Conservation Reserve Enhancement Program (CREP)
grasslands, Saginaw Bay watershed, Michigan, 2005-2006........................................

152

Table 3.10. Summary of AIC occupancy model selection adjusted by the best
detection model for Clay-colored Sparrow in native/CP23 vegetation of
Conservation Reserve Enhancement Program (CREP) grasslands, Saginaw Bay
watershed, Michigan, 2006..........................................................................................

153

Table 3.11. Best fitting single occupancy model for Clay-colored Sparrow
(Native/CP23 2006 data only), Eastern Kingbird, and Tree Swallow in CP23 and
CP1 Conservation Reserve Enhancement Program (CREP) grasslands, Saginaw
Bay watershed, Michigan, 2005-2006.........................................................................

154

Table 3.12. Summary of AIC occupancy model selection adjusted by the best
detection model for Eastern Kingbird in CP23 and CP1 CREP grasslands, Saginaw
Bay watershed, Michigan, 2006...................................................................................

155

Table 3.13. Summary of AIC occupancy model selection adjusted by the best
detection model for Tree Swallow in CP23 and CP1 Conservation Reserve
Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan,
2005-2006....................................................................................................................

156

xi

Appendix 3.2. Avian species observed in CP23 native fields, the introduced
portion of CP1 fields, and CP1 whole fields, Conservation Reserve Enhancement
Program (CREP) grasslands, Saginaw Bay watershed, Michigan, 2005-2006.
Avian densities (# individuals /ha [SE]) are given for species with detection
probability estimates ≥0.10. Species observed too rarely to support detection
probability modeling (i.e., observed in < 10% of surveys) or with detection
probability estimates <0.10 are noted as present when observed in at least one
survey and as undetected when unobserved................................................................

145

Table 4.1. All potential grassland bird occupancy covariates for CP23 native and
CP1 introduced plant communities of Conservation Reserve Enhancement Program
(CREP) grasslands, Saginaw Bay watershed, Michigan, 2005-2006..........................

188

Table 4.2. Potential occupancy covariates for grassland bird species observed in
CP23 native and CP1 introduced plant communities of Conservation Reserve
Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan,
2005-2006....................................................................................................................

190

Table 4.3. Grassland bird species, with corresponding best-fitting detection models
(with constant occupancy, ψ[·]) and detection probability estimates, observed in
CP23 native and CP1 introduced plant communities of Conservation Reserve
Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan,
2005-2006. The lowest detection probability estimate was reported when multiple
detection probability estimates were generated...........................................................

191

Table 4.4. Grassland bird species and global occupancy models with
corresponding number of parameters, K, and overdispersion parameters, ĉ,
observed in CP23 native and CP1 introduced plant communities of Conservation
Reserve Enhancement Program (CREP) grasslands, Saginaw Bay watershed,
Michigan, 2005-2006...................................................................................................

192

Table 4.5. Summary of AIC occupancy model selection adjusted by the best
detection model for Dickcissel in CP23 native and CP1 introduced plant
communities of Conservation Reserve Enhancement Program (CREP) grasslands,
Saginaw Bay watershed, Michigan, 2005-2006...........................................................

193

Table 4.6. Best fitting single occupancy models for Dickcissel and Grasshopper
Sparrow and composite (i.e., model averaged) occupancy models for Sedge Wren
(2006 data only) and Vesper Sparrow in CP23 native and CP1 introduced plant
communities of Conservation Reserve Enhancement Program (CREP) grasslands,
Saginaw Bay watershed, Michigan, 2005-2006...........................................................

195

Table 4.7. Summary of AIC occupancy model selection adjusted by the best
detection model for Eastern Meadowlark in CP23 native and CP1 introduced plant
communities of Conservation Reserve Enhancement Program (CREP) grasslands,
Saginaw Bay watershed, Michigan, 2005-2006...........................................................

196

xii

Table 4.8. Summary of AIC occupancy model selection adjusted by the best
detection model for Grasshopper Sparrow in CP23 native and CP1 introduced plant
communities of Conservation Reserve Enhancement Program (CREP) grasslands,
Saginaw Bay watershed, Michigan, 2005-2006...........................................................

198

Table 4.9. Summary of AIC occupancy model selection adjusted by the best
detection model for Ring-necked Pheasant in CP23 native and CP1 introduced
plant communities of Conservation Reserve Enhancement Program (CREP)
grasslands, Saginaw Bay watershed, Michigan...........................................................

200

Table 4.10. Summary of AIC occupancy model selection adjusted by the best
detection model for Sedge Wren in CP23 native and CP1 introduced plant
communities of Conservation Reserve Enhancement Program (CREP) grasslands,
Saginaw Bay watershed, Michigan.............................................................................

201

Table 4.11. Summary of AIC occupancy model selection adjusted by the best
detection model for Vesper Sparrow in CP23 native and CP1 introduced plant
communities of Conservation Reserve Enhancement Program (CREP) grasslands,
Saginaw Bay watershed, Michigan..............................................................................

202

Appendix 4.2. Avian species observed in CP23 native fields, the introduced
portion of CP1 fields, and CP1 whole fields, Conservation Reserve Enhancement
Program (CREP) grasslands, Saginaw Bay watershed, Michigan, 2005-2006.
Avian densities (# individuals /ha [SE]) are given for species with detection
probability estimates ≥0.10. Species observed too rarely to support detection
probability modeling (i.e., observed in < 10% of surveys) or with detection
probability estimates <0.10 are noted as present when observed in at least one
survey and as undetected when unobserved................................................................

204

xiii

LIST OF FIGURES

Fig. 1.1. Location, type, and number of Conservation Reserve Enhancement
Program (CREP) study fields within Tuscola County, Saginaw Bay watershed,
Michigan, 2005 and 2006. Wetlands and wetland buffers were excluded from avian
surveys and vegetation sampling. The surveyed (i.e., upland) area of CP23 fields
ranged in size from 6.9 – 19.8 ha (median [SE] = 12.6 [1.9] ha, median = 11.8 ha).
The introduced portion of CP1 fields ranged in size from 5.5 – 17.0 ha (mean [SE]
= 9.4 [1.5], median = 8.0 ha), and CP1 whole fields ranged in size from 7.3 – 23.9
ha (mean [SE] = 12.3 [2.1] ha, median = 10.9 ha........................................................

9

Fig. 2.1. Location, type, and number of Conservation Reserve Enhancement
Program (CREP) study fields within Tuscola County, Saginaw Bay watershed,
Michigan, 2005 and 2006. Wetlands and wetland buffers were excluded from avian
surveys and vegetation sampling. The surveyed (i.e., upland) area of CP23 fields
ranged in size from 6.9 – 19.8 ha (median [SE] = 12.6 [1.9] ha, median = 11.8 ha).
The introduced portion of CP1 fields ranged in size from 5.5 – 17.0 ha (mean [SE]
= 9.4 [1.5], median = 8.0 ha), and CP1 whole fields ranged in size from 7.3 – 23.9
ha (mean [SE] = 12.3 [2.1] ha, median = 10.9 ha). .....................................................

46

Fig. 2.2. 2005 mid-summer (15 June – 14 July) vegetation variables of native and
introduced vegetation of Conservation Reserve Enhancement Program (CREP)
grasslands, Saginaw Bay watershed, Michigan. For interpretation of the references
to color in this and all other figures, the reader is referred to the electronic version
of this dissertation........................................................................................................

50

Fig. 2.3. 2006 early summer (15 May - 14 June) vegetation variables of native and
introduced vegetation of Conservation Reserve Enhancement Program (CREP)
grasslands, Saginaw Bay watershed, Michigan. ..........................................................

51

Fig. 2.4. 2006 mid- summer (15 June - 14 July) vegetation variables of native and
introduced vegetation of Conservation Reserve Enhancement Program (CREP)
grasslands, Saginaw Bay watershed, Michigan. ..........................................................

52

Fig. 2.5. 2006 late summer (15 July - 14 August) vegetation variables of native and
introduced vegetation of Conservation Reserve Enhancement Program (CREP)
grasslands, Saginaw Bay watershed, Michigan. ..........................................................

53

Fig. 2.6. Native vegetation variables of CP23 and CP1 Conservation Reserve
Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan,
mid-summer (15 June – 14 July) 2005.........................................................................

54

Fig. 2.7. Native vegetation variables of CP23 and CP1 Conservation Reserve
Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan,
early summer (15 May – 14 June) 2006......................................................................

55

xiv

Fig. 2.8. Native vegetation variables of CP23 and CP1 Conservation Reserve
Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan,
mid-summer (15 June – 14 July) 2006.........................................................................

56

Fig. 2.9. Native vegetation variables of CP23 and CP1 Conservation Reserve
Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan, late
summer (15 July – 14 August) 2006............................................................................

57

Fig. 2.10. Relationship between Bobolink detection probability and ordinal date in
CP23 and CP1 Conservation Reserve Enhancement Program (CREP) grasslands,
Saginaw Bay watershed, Michigan, 2005-2006. Error bars indicate 95%
confidence intervals.....................................................................................................

62

Fig. 2.11. Relationship between Eastern Meadowlark detection probability and
Conservation Reserve Enhancement Program (CREP) administrative field type in
CP23 and CP1 CREP grasslands, Saginaw Bay watershed, Michigan, 2005 and
2006. Error bars indicate 95% confidence intervals....................................................

63

Fig. 2.12. Relationship between Ring-necked Pheasant detection probability, field
type, and ordinal date in CP23 and CP1 Conservation Reserve Enhancement
Program (CREP) grasslands, Saginaw Bay watershed, Michigan, 2005 and 2006.
Error bars indicate 95% confidence intervals..............................................................

64

Fig. 2.13. Relationship between Savannah Sparrow detection probability and
ordinal date in CP23 and CP1 Conservation Reserve Enhancement Program
(CREP) grasslands, Saginaw Bay watershed, Michigan, 2005-2006. Error bars
indicate 95% confidence intervals...............................................................................

65

Fig. 2.14. Relationship between Sedge Wren detection probability and ordinal date
in CP23 and CP1 Conservation Reserve Enhancement Program (CREP) grasslands,
Saginaw Bay watershed, Michigan, 2006. Error bars indicate 95% confidence
intervals........................................................................................................................

66

Fig. 2.15. Relationship between Dickcissel detection probability and ordinal date in
CP23 native and CP1 introduced plant communities of Conservation Reserve
Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan,
2005-2006. Error bars indicate 95% confidence intervals..........................................

67

Fig. 2.16. Relationship between Sedge Wren detection probability and grassland
type in CP23 native and CP1 introduced plant communities of Conservation
Reserve Enhancement Program (CREP) grasslands, Saginaw Bay watershed,
Michigan, 2005 and 2006. Error bars indicate 95% confidence intervals...................

68

Fig. 2.17. Savannah Sparrow density (# individuals / ha) in native/CP23 whole
fields, introduced vegetation of CP1 fields, and CP1 whole fields of Conservation
Reserve Enhancement Program (CREP) grasslands in Michigan‘s Saginaw Bay
watershed, 2005 and 2006............................................................................................

69

xv

Fig. 2.18. Bobolink density (# individuals / ha) in native/CP23 whole fields,
introduced vegetation of CP1 fields, and CP1 whole fields of Conservation Reserve
Enhancement Program (CREP) grasslands in Michigan‘s Saginaw Bay watershed,
2005 and 2006..............................................................................................................

70

Fig. 2.19. Dickcissel density (# individuals / ha) in native/CP23 whole fields,
introduced vegetation of CP1 fields, and CP1 whole fields of Conservation Reserve
Enhancement Program (CREP) grasslands in Michigan‘s Saginaw Bay watershed,
2005 and 2006..............................................................................................................

71

Fig. 2.20. Eastern Meadowlark density (# individuals / ha) in native/CP23 whole
fields, introduced vegetation of CP1 fields, and CP1 whole fields of Conservation
Reserve Enhancement Program (CREP) grasslands in Michigan‘s Saginaw Bay
watershed, 2005 and 2006............................................................................................

72

Fig. 2.21. Grasshopper Sparrow density (# individuals / ha) in native/CP23 whole
fields, introduced vegetation of CP1 fields, and CP1 whole fields of Conservation
Reserve Enhancement Program (CREP) grasslands in Michigan‘s Saginaw Bay
watershed, 2005 and 2006............................................................................................

73

Fig. 2.22. Ring-necked Pheasant density (# individuals / ha) in native/CP23 whole
fields, introduced vegetation of CP1 fields, and CP1 whole fields of Conservation
Reserve Enhancement Program (CREP) grasslands in Michigan‘s Saginaw Bay
watershed, 2005 and 2006............................................................................................

74

Fig. 2.23. Sedge Wren density (# individuals / ha) in native/CP23 whole fields,
introduced vegetation of CP1 fields, and CP1 whole fields of Conservation Reserve
Enhancement Program (CREP) grasslands in Michigan‘s Saginaw Bay watershed,
2005 and 2006..............................................................................................................

75

Fig. 2.24. Vesper Sparrow density (# individuals / ha) in native/CP23 whole fields,
introduced vegetation of CP1 fields, and CP1 whole fields of Conservation Reserve
Enhancement Program (CREP) grasslands in Michigan‘s Saginaw Bay watershed,
2005 and 2006..............................................................................................................

76

Fig. 2.25. American Goldfinch density (# individuals / ha) in native/CP23 whole
fields, introduced vegetation of CP1 fields, and CP1 whole fields of Conservation
Reserve Enhancement Program (CREP) grasslands in Michigan‘s Saginaw Bay
watershed, 2005 and 2006............................................................................................

85

Fig. 2.26. Common Yellowthroat density (# individuals / ha) in native/CP23 whole
fields, introduced vegetation of CP1 fields, and CP1 whole fields of Conservation
Reserve Enhancement Program (CREP) grasslands in Michigan‘s Saginaw Bay
watershed, 2005 and 2006............................................................................................

86

xvi

Fig. 2.27. Eastern Kingbird density (# individuals / ha) in native/CP23 whole fields
and introduced vegetation of CP1 fields of Conservation Reserve Enhancement
Program (CREP) grasslands in Michigan‘s Saginaw Bay watershed, 2006................

87

Fig. 2.28. Red-winged blackbird density (# individuals / ha) in native/CP23 whole
fields, introduced vegetation of CP1 fields, and CP1 whole fields of Conservation
Reserve Enhancement Program (CREP) grasslands in Michigan‘s Saginaw Bay
watershed, 2005 and 2006............................................................................................

88

Fig. 2.29. Song Sparrow density (# individuals / ha) in native/CP23 whole fields,
introduced vegetation of CP1 fields, and CP1 whole fields of Conservation Reserve
Enhancement Program (CREP) grasslands in Michigan‘s Saginaw Bay watershed,
2005 and 2006..............................................................................................................

89

Fig. 2.30. Tree Swallow density (# individuals / ha) in native/CP23 and whole
fields of Conservation Reserve Enhancement Program (CREP) grasslands in
Michigan‘s Saginaw Bay watershed, 2006..................................................................

90

Fig. 3.1. Location, type, and number of Conservation Reserve Enhancement
Program (CREP) study fields within Tuscola County, Saginaw Bay watershed,
Michigan, 2005 and 2006. Wetlands and wetland buffers were excluded from avian
surveys and vegetation sampling. The surveyed (i.e., upland) area of CP23 fields
ranged in size from 6.9 – 19.8 ha (median [SE] = 12.6 [1.9] ha, median = 11.8 ha).
The introduced portion of CP1 fields ranged in size from 5.5 – 17.0 ha (mean [SE]
= 9.4 [1.5], median = 8.0 ha), and CP1 whole fields ranged in size from 7.3 – 23.9
ha (mean [SE] = 12.3 [2.1] ha, median = 10.9 ha).......................................................

131

Fig. 3.2. Relationship between Bobolink detection probability and ordinal date in
CP23 and CP1 Conservation Reserve Enhancement Program (CREP) grasslands,
Saginaw Bay watershed, Michigan, 2005-2006. Error bars indicate 95%
confidence intervals......................................................................................................

133

Fig. 3.3. Relationship between Eastern Meadowlark detection probability and
Conservation Reserve Enhancement Program (CREP) administrative field type in
CP23 and CP1 CREP grasslands, Saginaw Bay watershed, Michigan, 2005 and
2006. Error bars indicate 95% confidence intervals....................................................

134

Fig. 3.4. Relationship between Ring-necked Pheasant detection probability, field
type, and ordinal date in CP23 and CP1 Conservation Reserve Enhancement
Program (CREP) grasslands, Saginaw Bay watershed, Michigan, 2005 and 2006.
Error bars indicate 95% confidence intervals...............................................................

135

Fig. 3.5. Relationship between Savannah Sparrow detection probability and ordinal
date in CP23 and CP1 Conservation Reserve Enhancement Program (CREP)
grasslands, Saginaw Bay watershed, Michigan, 2005-2006. Error bars indicate
95% confidence intervals.............................................................................................

136

xvii

Fig. 3.6. Relationship between Sedge Wren detection probability and ordinal date
in CP23 and CP1 Conservation Reserve Enhancement Program (CREP) grasslands,
Saginaw Bay watershed, Michigan, 2006. Error bars indicate 95% confidence
intervals........................................................................................................................

137

Fig. 4.1. Location, type, and number of Conservation Reserve Enhancement
Program (CREP) study fields within Tuscola County, Saginaw Bay watershed,
Michigan, 2005 and 2006. Wetlands and wetland buffers were excluded from avian
surveys and vegetation sampling. The surveyed (i.e., upland) area of CP23 fields
ranged in size from 6.9 – 19.8 ha (median [SE] = 12.6 [1.9] ha, median = 11.8 ha).
The introduced portion of CP1 fields ranged in size from 5.5 – 17.0 ha (mean [SE]
= 9.4 [1.5], median = 8.0 ha), and CP1 whole fields ranged in size from 7.3 – 23.9
ha (mean [SE] = 12.3 [2.1] ha, median = 10.9 ha.........................................................

185

Fig. 4.2. Relationship between Dickcissel detection probability and ordinal date in
CP23 native and CP1 introduced plant communities of Conservation Reserve
Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan,
2005-2006. Error bars indicate 95% confidence intervals..........................................

186

Fig. 4.3. Relationship between Sedge Wren detection probability and grassland
type in CP23 native and CP1 introduced plant communities of Conservation
Reserve Enhancement Program (CREP) grasslands, Saginaw Bay watershed,
Michigan, 2005 and 2006. Error bars indicate 95% confidence
intervals........................................................................................................................

187

xviii

CHAPTER 1: INTRODUCTION, RESEARCH OBJECTIVES,
AND STUDY AREA

INTRODUCTION
Native temperate grassland, also called prairie or steppe, is one of the most
endangered ecosystems in the world (Sampson and Knopf 1996). Vast areas have been
converted to agriculture, and temperate grassland currently has the lowest protected area
relative to converted area of any major biome (Hoekstra et al. 2005). Remaining
grasslands tend to be fragmented and isolated, which disrupts natural ecological processes
and reduces the viability of the grassland patches that persist (Leach and Givnish 1996).
Other significant and interacting threats to grassland persistence, quality, and extent include
changes to fire intensity and frequency, incompatible grazing regimes, woody plant
expansion, and climate change (Briggs et al. 2005). In North America, as much as 99.9%
of native prairies have vanished since European settlement (Sampson and Knopf 1994),
and in the American Midwest region, 83-99% of native grasslands have been lost (Noss et
al. 1995).
The decline of native North American grasslands has had significant negative
impacts on associated animal species (Arenz and Joern 1996, Benedict et al. 1996, Corn
and Peterson 1996, Rabeni 1996). In particular, a decrease in grassland bird populations
has been documented (Knopf 1994, Herkert 1995). Grassland birds include approximately
30 species that evolved in the pre-European settlement grassland landscape of the North
American Great Plains (Sauer et al. 1995, Knopf 1996). Breeding Bird Survey (BBS) data
indicate that grassland birds, particularly endemic species, have declined more rapidly,

1

more consistently, and over a wider geographic area than any other guild of North
American birds (Sauer et al. 1995, Knopf 1996).
Farmland set-aside programs such as the Conservation Reserve Enhancement
Program (CREP) increase the number of native and non-native grasslands in the U.S. and
regulate management practices on enrolled private lands (e.g., mowing occurrence and
timing), which may help diminish or reverse the decline of grassland bird species (Herkert
2009). CREP, a federal farmland set-aside program authorized by the 1996 Farm Bill and
based upon another farmland set-aside program, the Conservation Reserve Program
(CRP), has goals of protecting environmentally sensitive land, controlling soil erosion,
safeguarding ground and surface water quality, and restoring wildlife habitat (USDA
2009, FSA 2010a). In the U. S., 10,805,106 ha are currently enrolled in CRP, and
481,576 ha are enrolled in CREP (FSA 2010b). In Michigan, 94,156 ha are currently
enrolled in CRP, and 28,152 ha are enrolled in CREP (USDA 2010).
Both CREP and CRP are administered by the U.S. Department of Agriculture‘s
(USDA) Farm Service Agency (FSA) and provide cost-share assistance and annual rental
payments to farmers that remove highly erodible, environmentally sensitive cropland
from agricultural production and plant these set aside fields with grasses, trees, or other
approved vegetation under 10- to 15-year contracts (USDA 2009, FSA 2010a).
However, while CRP is strictly a federal program, CREP is a partnership between federal
and state governments, with additional partners in some cases, including tribal
governments and private groups (FSA 2010a). CREP also differs from CRP in that, to
address high-priority conservation areas, enrollment is permitted only in select
geographic areas within participating states (FSA 2010a). Also, CREP offers greater

2

financial incentives to eligible private landowners than CRP by leveraging CRP funding
against state, tribal and other non-federal funds (FSA 2010a).
In Michigan, CREP is administered by the Michigan Department of Agriculture
(MDA), with support from local Soil and Water Conservation Districts. The focus of the
program is restricted to the River Raisin, Lake Macatawa, and Saginaw Bay watersheds
(MDA 2009). The CREP conservation practice that is the focus of this study is grass
plantings, non-native grasses and legumes or native grasses and wildflowers planted in
fields to reduce erosion of cropland and provide wildlife habitat (MDA 2009). Other
CREP conservation practices used in Michigan include: 1) filter strips, which are grass
and legume plantings in strips adjacent to waterways or at the lower edges of fields to
trap sediments and pollution in rain or snowmelt runoff and to provide wildlife habitat; 2)
riparian buffers, or plantings of trees, shrubs, and grasses at the edges of streams or drains
to filter runoff and provide wildlife habitat; 3) field windbreaks, or plantings of trees or
shrubs to reduce wind erosion, protect crops, and provide wildlife habitat; and 4) wetland
restoration in lands with hydric soils to improve water quality, recharge aquifers, provide
flood control, and provide wildlife habitat (MDA 2009).
Geographic variation in the effects of set-aside lands on wildlife exists due to local
and regional climatic differences and variation in program implementation, including
management practices (Heard et al. 2000, Riffell et al. 2008). As a result, it is important to
assess the impacts of farmland set-aside programs on wildlife at sub-national scales, such
as state or major watershed levels.
In Michigan, at least 39 tallgrass prairies covered approximately 1 million ha prior
to European settlement (Sargent and Carter 1999). Fewer than 810 ha of these prairies, or

3

less than 0.1%, remain (Sargent and Carter 1999). Six grassland bird species are among
those that have been designated as Michigan priority land birds by the bird conservation
group Partners in Flight (Knutson et al. 2001, Matteson et al. 2009) and have been given
management priority by the Michigan Department of Natural Resources (MDNR). These
species are Bobolink, Dickcissel, Grasshopper Sparrow, Henslow‘s Sparrow, Northern
Harrier, and Sedge Wren (Knutson et al. 2001, Matteson et al. 2009). Dickcissel,
Grasshopper Sparrow, and Northern Harrier are also state species of special concern, and
Henslow‘s Sparrow is a state endangered species. Several studies have evaluated the
impacts of CRP on bird communities in Michigan (e.g., Pearks 1995, Millenbah et al.
1996, Minnis 1996, Best et al. 1997, Best et al. 1998). Currently, no studies have
evaluated the vegetation characteristics of CREP grasslands in Michigan or their
suitability as habitat for grassland bird species.
RESEARCH OBJECTIVES
The goal of this study was to evaluate grassland bird responses to the administration
of CREP in Michigan. Objectives included:
1. describe and compare the structure and composition of native and introduced plant
communities in CREP grasslands (Chapter 2);
2. model detection probability for avian species observed in CREP grasslands to inform
future survey designs and identify those species that could reliably be included in
more extensive data analyses (Chapters 2, 3, and 4);
3. compare avian density for species with adequate detection probability estimates (i.e.,
p ≥ 0.10) in:

4

a. CREP Conservation Practice 23 (CP23) and Conservation Practice 1 (CP1)
whole fields (administrative-level comparison) (Chapter 2) and
b.

native grassland plant communities of CP23 fields and introduced grassland
plant communities of CP1 grassland plant communities (ecological-level
comparison) (Chapter 2);

4. model occupancy for grassland bird species with adequate detection probability
estimates in:
a. CP23 and CP1 CREP whole fields (administrative-level comparison)
(Chapter 3); and
b. native and introduced portions of CREP grassland plant communities
(ecological-level comparison) (Chapter 4) and
5. for grassland bird species with adequate detection probability estimates, evaluate the
correlation of occupancy with structural variables of native and introduced plant
communities and/or proximate environmental cues (e.g., area of woody vegetation
≤100 m from study areas) (Chapter 4).
STUDY AREA
2

This study occurred within the 8,219 km Sandusky Lake Plain sub-subsection of
Michigan‘s southern Lower Peninsula (Albert 1995). This sub-subsection consists of flat
clay lake plain, bordered by shoreline dunes and sand plain. In the center of the flat clay
lake plain, long and narrow till plains and ridges of end moraines parallel either Saginaw
Bay or Lake Huron (Albert 1995). Prior to European settlement, extensive wet and wetmesic prairies occurred upland from the coastal Great Lakes marshes of Saginaw Bay, but
now occur only as small remnants, generally on State-owned lands (Albert 1995).

5

Other presettlement vegetation types of the Sandusky Lake Plain sub-subsection
differed between the clay lake plain, end moraine, and sand lake plain (Albert 1995). The
vegetation of the clay lake plain was dominated by hemlock (Tsuga canadensis) upland
conifer forests, and included lowland hardwoods in the wettest areas and beech (Fagus
grandifolia)- sugar maple (Acer saccharum) forest in more well-drained areas (Albert
1995). Beech - sugar maple forests were also typical of the sloping end moraines, with
conifer swamps occupying the broad moraine footslope wetlands (Albert 1995). The flat
sand plain and low dunes supported hemlock - white pine (Pinus strobus) conifer forests
and small areas of white oak (Quercus alba) - black oak (Quercus velutina) forest (Albert
1995). Along the shores of Saginaw Bay, extensive dune - swale wetland complexes and
emergent marshes occurred (Albert 1995). Intensive agricultural development has occurred
throughout the sub-subsection due to productive loamy soils and a lake-moderated climate
(Albert 1995). Most of the clay lake plain, parts of the sand plain, the moraines, and large
areas of wet prairie and marsh have been converted to farmland, leaving few presettlement
cover types in the Sandusky Lake Plain sub-subsection (Albert 1995).
In Michigan, 22,983 ha of privately-owned land in the Saginaw Bay, Lake
Macatawa, and River Raisin watersheds were enrolled in CREP at the time of this study
(USDA 2007). From CREP lands in the Saginaw Bay watershed, 14 CREP grassland
fields in Tuscola County, ranging in size from 6.8 to 23.6 ha, were selected for study
(Figure 1.1). All grassland study fields were planted in 2002 and were surveyed for birds
in 2005 and 2006. Seven study fields were CREP conservation practice (CP) type 23
(hereafter CP23) that had been planted with a seed mixture of native grasses and native
and introduced wildflowers over 100% of the field area, and 7 study fields were CP type

6

1 (hereafter CP1) fields that had been planted with introduced grasses and legumes over
70% of the field area, with the remaining 30% of the field area planted with the same
native seed mixture used in CP23 fields (Figure 1.1, Table 1.1). Native sections of CP1
fields were established to provide winter shelter for wildlife (e.g., Ring-necked Pheasant,
Phasianus colchicus) because native grasses tend to remain upright during winter
whereas introduced grasses often collapse (USDA 2000a, USDA 2000b). In two CP1
study fields, the native seed mixture did not successfully establish and the field area was
composed only of introduced vegetation (Figure 1.1). In 2005, the native seed mixture
cost $16.00/pound (i.e., $104.00/acre or $256.99/hectare) and the introduced seed
mixture cost $1.90/pound (i.e., $20.90/acre or $51.65/hectare) plus 6% sales tax through
the Tuscola County Conservation District.
Within CREP, CP23 fields are wetland restorations and all CP23 fields had hydric
soils. A small wetland (i.e., < 10% of upland area) was created in each CP23 field by
cutting drain tiles, and if necessary, by creating a push-out, or shallow depression in the
soil. The CREP program offers a whole-field native grassland conservation practice
without a wetland component, CP2, which was rare in the Saginaw Bay area. Due to the
scarcity of CP2 fields, upland areas of CP23 fields were used for comparison to 1) CP1
whole fields and 2 introduced portions of CP1 fields). Wetlands and buffers of 25 m
surrounding each wetland were not included in CP23 avian or vegetation surveys. The
surveyed (i.e., upland) area of CP23 fields ranged in size from 6.9 – 19.8 ha (median [SE]
= 12.6 [1.9] ha, median = 11.8 ha). CP1 whole fields ranged in size from 7.3 – 23.9 ha
(mean [SE] = 12.3 [2.1] ha, median = 10.9 ha), and the introduced portion of CP1 fields
ranged in size from 5.5 – 17.0 ha (mean [SE] = 9.4 [1.5], median = 8.0 ha).

7

APPENDIX 1.1: CHAPTER 1 TABLES AND FIGURES

8

CP1 Field
(standard)
n=5

CP23 Field
n=7

Native
Vegetation

Example
Wetland and
25 m Wetland
Buffer

Introduced
Vegetation
(70% area)

CP1 Field
(native planting failed)
n=2
Native
Vegetation
(30% area)

Introduced
Vegetation

Fig. 1.1. Location, type, and number of Conservation Reserve Enhancement Program
(CREP) study fields within Tuscola County, Saginaw Bay watershed, Michigan, 2005
and 2006. Wetlands and wetland buffers were excluded from avian surveys and
vegetation sampling. The surveyed (i.e., upland) area of CP23 fields ranged in size from
6.9 – 19.8 ha (median [SE] = 12.6 [1.9] ha, median = 11.8 ha). The introduced portion of
CP1 fields ranged in size from 5.5 – 17.0 ha (mean [SE] = 9.4 [1.5], median = 8.0 ha),
and CP1 whole fields ranged in size from 7.3 – 23.9 ha (mean [SE] = 12.3 [2.1] ha,
median = 10.9 ha).

9

Table 1.1. Plant species and proportions of species included in the seed mixtures used to
establish the native and introduced vegetation communities of Conservation Reserve
Enhancement Program (CREP) grasslands of Michigan‘s Saginaw Bay watershed, 2002.
% Grass,
% Total Wildflower,
Seed
or Legume
Common Name
Scientific Name
Mixture
Mixture
CP23 Fields and Native Section of CP1 Fields - planted 6.5 pounds/acre (16.1 pounds/ha)
Native Grasses
92.3
Big Bluestem*
Andropogon gerardii
33.3
Indiangrass*
Sorghastrum nutans
33.3
Little Bluestem*
Andropogon scoparius
33.3
Plateau® Tolerant Wildflower Mixture
7.7
Black Eyed Susan*
Rudbeckia hirta
10.0
Blanket Flower, Perennial
Gaillardia aristata
2.0
Catchfly
Silene armeria
2.0
Clover, Crimson
Trifolim incarnatum
10.0
Coneflower, Prairie, 'Mexican Hat'
Ratibida columnifera
3.0
Coneflower, Purple*
Echinacea purpurea
2.0
Coneflower, Yellow*
Ratibida columnifera
5.0
Coreopsis, Lance Leaf*
Coreopsis lanceolata
10.0
Coreopsis, Plains
Coreopsis tinctoria
3.0
Cosmos, ‗Sensation‘ Mix
Cosmos bipinnatus
5.0
Cosmos, Sulphur
Cosmos sulphureus
5.0
Daisy, Shasta, 'Alaska'
Chrysanthemum x superbum
4.0
Dame‘s Rocket
Hesperis matronalis
5.0
Flax, Perennial
Linum perenne lewisii
10.0
Lupine, Perennial*
Lupinus perennis
4.0
Poppy, California
Eschscholzia californica
5.0
Poppy, Corn, ‗Shirley Single‘ Mix Papaver rhoeas
5.0
Wallflower, Siberian
Erysiumum x marshallii
10.0
Introduced Section of CP1 Fields - planted 11 pounds/acre (27.2 pounds/ha)
Introduced Grasses
45.5
Orchardgrass
Dactlyis glomerata
50.0
Timothy
Phleum pratense
50.0
Introduced Legumes
54.5
Alfalfa
Medicago sativa
50.0
Red Clover
Trifolium pratense
50.0
*Michigan native species (Voss 1985, Voss 1996, USDA NRCS 2011)

10

CHAPTER 1 LITERATURE CITED

11

Albert, D. A. 1995. Regional landscape ecosystems of Michigan, Minnesota, and
Wisconsin: a working map and classification. General Technical Report NC-178.
St. Paul, Minnesota, USA: U.S. Department of Agriculture, Forest Service, North
Central Forest Experiment Station.

Arenz, C. L., and A. Joern. 1996. Prairie legacies - invertebrates. Pages 91-109 in F. B.
Samson and F. L. Knopf, editors. Prairie conservation: preserving North
America‘s most endangered ecosystem. Island Press, Covelo, California, USA.

Benedict, R. A., P.W. Freeman, and H. H. Genoways. 1996. Prairie legacies - mammals.
Pages 149-166 in F. B. Samson and F. L. Knopf, editors. Prairie conservation:
preserving North America‘s most endangered ecosystem. Island Press, Covelo,
California, USA.

Best, L. B., H. Campa, III, K. E. Kemp, R. J. Robel, M. R. Ryan, J. A. Savidge, H. P.
Weeks, Jr., and S. R. Winterstein. 1997. Bird abundance and nesting in CRP
fields and cropland in the Midwest. Wildlife Society Bulletin 28:490-495.

Best, L. B., H. Campa, III, K. E. Kemp, R. J. Robel, M. R. Ryan, J. A. Savidge, H. P.
Weeks, Jr., and S. R. Winterstein. 1998. Avian abundance in CRP and crop
fields during winter months in the Midwest. American Midland Naturalist
139:311-324.

Briggs, J. M., A. K. Knapp, J. M. Blair, J. L Heisler, G. A. Hoch, M. S. Lett, and J. K.
McCarron. 2005. An ecosystem in transition: causes and consequences of the
conversion of mesic grassland to shrubland. BioScience 55:243-254.
Corn, P. S., and C. R. Peterson. 1996. Prairie legacies – amphibians and reptiles. Pages
125-134 in F. B. Samson and F. L. Knopf, editors. Prairie conservation:
preserving North America‘s most endangered ecosystem. Island Press, Covelo,
California, USA.

FSA (Farm Service Agency). 2010a. Conservation Reserve Enhancement Program.
http://www.apfo.usda.gov/FSA/webapp?area=home&subject=copr&topic=cep.
Accessed June 2010.

12

FSA (Farm Service Agency). 2010b. Conservation Reserve Program status April 30,
2010. http://www.fsa.usda.gov/Internet/FSA_File/april2010onepager.pdf.
Accessed June 2010.

Heard, L. P., A. W. Allen, L. B. Best, S. J. Brady, W. Burger, A. J. Esser, E. Hackett, D. H.
Johnson, R. L. Pederson, R. E. Reynolds, C. Rewa, M. R. Ryan, R. T. Molleur, and
P. Buck. 2000. A comprehensive review of Farm Bill contributions to wildlife
conservation, 1985-2000. W. L. Hohman, and D. J. Halloum, editors. U.S.D.A.,
Natural Resources Conservation Service, Wildlife Habitat Management Institute,
Technical Report USDA/NRCS/WHMI-2000.

Herkert, J. R. 1995. An analysis of Midwestern breeding bird population trends: 19661993. American Midland Naturalist 134:41-50.

Herkert, J. R. 2009. Response of bird populations to farmland set-aside programs.
Conservation Biology 23:1036-1040.

Hoekstra, J. M., T. M. Boucher, T. H. Ricketts, and C. Roberts. 2005. Confronting a
biome crisis: global disparities of habitat loss and protection. Ecology Letters 9:2329.

Knopf, F. L. 1994. Avian assemblages on altered grasslands. Studies in Avian Biology
15:247-257.

Knopf, F. L. 1996. Prairie legacies - birds. Pages 115-141 in F. B. Samson and F. L.
Knopf, editors. Prairie conservation: preserving North America‘s most endangered
ecosystem. Island Press, Covelo, California, USA.

Knutson, M. G., G. Butcher, J. Fitzgerald, and J. Shieldcastle. 2001. Partners in Flight
Bird Conservation Plan for The Upper Great Lakes Plain (Physiographic Area
16). USGS Upper Midwest Environmental Sciences Center in cooperation with
Partners in Flight. La Crosse, Wisconsin, USA.

Leach, M. K., and T. J. Givnish. 1996. Ecological determinants of species loss in remnant
prairies. Science 273:1555-1558.

13

Matteson, S., K. Kreitinger, G. Bartelt, G. Butcher, D. Sample, and T. Will. 2009.
Partners in Flight Bird Conservation Plan for The Boreal Hardwood Transition
(Bird Conservation Region 12 — U.S. Portion). Version 1.0. Partners in Flight.
www.partnersinflight.org. Accessed November 2010.

MDA (Michigan Department of Agriculture). 2009. What is CREP?
http://www.michigan.gov/mda/0,1607,7-125-1567_1599_1603-9735--,00.html.
Accessed June 2010.

Millenbah, K. F., S. R. Winterstein, H. R. Campa, III, L. T. Furrow, and R. B. Minnis.
1996. Effects of Conservation Reserve Program field age on avian relative
abundance, diversity, and productivity. Wilson Bulletin 108:760-770.

Minnis, R. B. 1996. Monitoring the effects of Conservation Reserve Program lands on
the diversity and abundance of wildlife and plant species in a temperate agroecosystem. M. S. Thesis, Michigan State University, East Lansing, Michigan,
U.S.A.

Noss, R. F., E. T. Laroe, and J. M. Scott. 1995. Endangered ecosystems of the United
States: a preliminary assessment of loss and degradation. National Biological
Service, U.S. Department of the Interior, Report No. 0611-R-01 (MF).
Washington, D. C., USA.

Pearks, A. J. 1995. The effects of habitat manipulation on vegetation characteristics and
avian communities on Conservation Reserve Program fields in Gratiot County,
Michigan. M. S. Thesis, Michigan State University, East Lansing, Michigan,
USA.
Rabeni, C. F. 1996. Prairie legacies – fish and aquatic resources. Pages 111-124 in F. B.
Samson and F. L. Knopf, editors. Prairie conservation: preserving North
America‘s most endangered ecosystem. Island Press, Covelo, California, USA.

Riffell, S., D. Scognamillo, and L.W. Burger. 2008. Effects of the Conservation Reserve
Program on northern bobwhite and grassland birds. Environmental Monitoring
and Assessment 146:309-323.

Sampson, F. B. and F. L. Knopf. 1994. Prairie conservation in North America.
BioScience 44:418-421.
14

Sampson, F. B., and F. L. Knopf, editors. 1996. Prairie conservation: preserving North
America‘s most endangered ecosystem. Island Press, Washington, D. C., USA.
Sargent, M. and K. S. Carter, editors. 1999. Managing Michigan‘s wildlife: a
landowner‘s guide. Michigan United Conservation Clubs, East Lansing,
Michigan, USA.

Sauer, J. R., B. G. Peterjohn, S. Schwartz, and J. E. Hines. 1995. The grassland bird home
page. Version 95.0. Patuxent Wildlife Research Center, Laurel, Maryland, USA.
http://www.mbr-pwrc.usgs.gov/bbs/grass/grass.htm. Accessed May 2010

USDA (U.S. Department of Agriculture). 2000a. Introduced grass planting: Conservation
Reserve Enhancement Program, CREP CP-1. Natural Resources Conservation
Service, Michigan. http://www.michigan.gov/documents/mda_ CP1_INTRODUCEDGRASS_9614_7.pdf. Accessed May 2010.

USDA (U.S. Department of Agriculture). 2000b. Native grass planting: Conservation
Reserve Enhancement Program, CREP CP-2. Natural Resources Conservation
Service, Michigan. http://www.michigan.gov/documents/mda_CP-2_
NATIVEGRASS_9615_7.pdf. Accessed May 2010.

USDA (U.S. Department of Agriculture). 2007. Conservation Reserve Program summary
and enrollment statistics, fiscal year 2006. Farm Service Agency.
http://www.fsa.usda.gov/Internet/FSA_File/06rpt.pdf. Accessed May 2010.

USDA (U.S. Department of Agriculture). 2009. Conservation Policy: Land Retirement
Programs. http://www.ers.usda.gov/Briefing/ConservationPolicy/ retirement.htm.
Accessed June 2010.

USDA (U.S. Department of Agriculture). 2010. Conservation Reserve Program monthly
summary April 2010. http://www.fsa.usda.gov/Internet/FSA_File/ apr2010.pdf.
Accessed June 2010.

USDA NRCS (U.S. Department of Agriculture Natural Resources Conservation Service).
2011. The PLANTS Database. National Plant Data Center, Baton Rouge,
Louisiana, USA. http://plants.usda.gov. Accessed January 2011.

15

Voss, E. G. 1985. Michigan Flora: a guide to the identification and occurrence of the
native and naturalized seed-plants of the state; Part II: Dicots (SaururaceaeCornaceae). Cranbrook Institute of Science Bulletin 59 and University of
Michigan Herbarium, Ann Arbor, Michigan, USA.

Voss, E. G. 1996. Michigan Flora: a guide to the identification and occurrence of the
native and naturalized seed-plants of the state; Part III: Dicots (PyrolaceaeCompositae). Cranbrook Institute of Science Bulletin 61 and University of
Michigan Herbarium, Ann Arbor, Michigan, USA.

16

CHAPTER 2: GRASSLAND BIRD DETECTION PROBABILITY ESTIMATION
AND DENSITY COMPARISON IN MICHIGAN CONSERVATION RESERVE
ENHANCEMENT PROGRAM (CREP) GRASSLANDS

INTRODUCTION
Temperate grassland is one of the most endangered ecosystems in the world,
having the lowest protected area relative to converted area of any major biome (Sampson
and Knopf 1996, Hoekstra et al. 2005). Vast areas of temperate grassland have been
converted to agriculture, and other significant and interacting threats to grassland
persistence, quality, and extent include changes to historical fire frequency and intensity,
incompatible grazing regimes, woody plant expansion, and climate change (Briggs et al.
2005). Remaining North American native grasslands (1–17% in the Midwest) tend to be
fragmented, which disrupts natural ecological processes and reduces the viability of
grassland patches that persist (Noss et al. 1995, Leach and Givnish 1996).
The decline of native North American grasslands has negatively impacted
associated animal species (Arenz and Joern 1996, Benedict et al. 1996, Corn and Peterson
1996, Rabeni 1996), and in particular, decreases in grassland bird populations have been
quantified (Knopf 1994, Herkert 1995). Breeding Bird Survey (BBS) data indicate that
grassland bird populations have declined more rapidly, more consistently, and over a
wider geographic area than any other guild of North American birds (Sauer et al. 1995,
Knopf 1996), and populations of grassland bird species continue to decline in Michigan
(Sauer et al. 2008). The most recent population trend analysis of Breeding Bird Survey
(BBS) data from 1980 – 2007 shows significantly decreasing population trends in

17

Michigan (p < 0.10) for 4 out of 10 grassland bird species included in the BBS analysis
(i.e., Vesper Sparrow, Eastern Meadowlark, Bobolink, and Ring-necked Pheasant;
species included in the analysis were those encountered on ≥ 15 survey routes), and no
species showed significant population increases (Sauer et al. 2008).
Farmland set-aside programs such as the Conservation Reserve Enhancement
Program (CREP) increase the number of native and non-native grasslands in the U.S.,
which may diminish or reverse the decline of grassland bird populations (Herkert 2009).
Geographic variation in the effects of farmland set-aside lands on wildlife exists due to
local and regional climatic differences and variation in program implementation,
including management practices (Heard et al. 2000, Riffell et al. 2008). As a result, it is
important to understand the impacts of farmland set-aside programs on wildlife at subnational scales, such as state or watershed levels. Currently, no studies have evaluated
the vegetation characteristics of CREP grasslands in Michigan or their suitability as
habitat for grassland bird species.
The goal of this study was to quantify the relationships between CREP grasslands
and grassland bird density in Michigan. In this study, CREP grasslands were identified
as CREP conservation practice (CP) 23 plantings (hereafter CP23 fields or native fields),
which were composed of native grasses and wildflowers, and CREP conservation
practice 1 plantings (hereafter CP1 fields or introduced fields), in which 70% of field area
consisted of introduced grasses and legumes and 30% of the field area was native grasses
and wildflowers. Study objectives were to: 1) describe and compare the vegetation
composition and structure of CP23 and CP1 fields and 2) describe and compare the
density of grassland bird species in a) CP23 and CP1 fields, corresponding to the scale at

18

which CREP is administered (hereafter referred to as the administrative level of inquiry)
and b) native and introduced CREP grassland plant communities (hereafter referred to as
the ecological level of inquiry).
METHODS
2

This study occurred within the 8,219 km Sandusky Lake Plain sub-subsection of
Michigan‘s southern Lower Peninsula (Albert 1995). This sub-subsection consists of flat
clay lake plain, bordered by shoreline dunes and sand plain. In the center of the flat clay
lake plain, long and narrow till plains and ridges of end moraines parallel either Saginaw
Bay or Lake Huron. Prior to European settlement, extensive wet and wet-mesic prairies
occurred upland from the coastal Great Lakes marshes of Saginaw Bay, but largely due to
agricultural conversion, these prairies now occur only as small remnants, generally on
State-owned lands.
In Michigan, 22,983 ha of land in the Saginaw Bay, Lake Macatawa, and River
Raisin watersheds were enrolled in CREP at the time of this study (USDA 2007). From
CREP lands in the Saginaw Bay watershed, 14 CREP grassland fields in Tuscola County,
ranging in size from 6.8 to 23.6 ha, were selected for study (Figure 2.1). All grassland
study fields were planted in 2002 and were surveyed for birds in 2005 and 2006. Seven
study fields were CP23 fields that had been planted with a seed mixture of wildflowers
and native grasses over 100% of the field area, and 7 study fields were CP1 fields that
had been planted with introduced grasses and legumes over 70% of the field area, with
the remaining 30% of the field area planted with the same native seed mixture used in
CP23 fields (Figure 2.1, Table 2.1). Native sections of CP1 fields were established to
provide winter shelter for wildlife (e.g., Ring-necked Pheasant, Phasianus colchicus)

19

because native grasses tend to remain upright during winter conditions whereas
introduced grasses generally collapse (USDA 2000a, 2000b). In two CP1 fields, the
native seed mixture did not successfully establish and the field area was composed only
of introduced vegetation (Figure 2.1). In 2005, the native seed mixture cost
$16.00/pound (i.e., $104.00/acre or $256.99/hectare) and the introduced seed mixture
cost $1.90/pound (i.e., $20.90/acre or $51.65/hectare) plus 6% sales tax through the
Tuscola County Conservation District.
Within CREP, CP23 fields are wetland restorations and all CP23 fields had hydric
soils. A small wetland was created in each CP23 field by cutting drain tiles, and, if
necessary, by creating a push-out, or shallow depression in the soil. Wetlands of the
CP23 fields were generally less than 3% of the upland field area, but in one field the
wetland was approximately 10% of the upland field area. The CREP program offers a
whole-field native grassland conservation practice without a wetland component, CP2,
which was rare in the Saginaw Bay area. Due to the scarcity of CP2 fields, the upland
area of CP23 fields was the native grassland used for comparison to the introduced
portion of CP1 fields and to CP1 whole fields. Wetlands and a buffer of 25 m
surrounding the wetlands were not included in CP23 avian surveys (Figure 2.1). The
surveyed (i.e., upland) area of CP23 fields ranged in size from 6.9 – 19.8 ha (median [SE]
= 12.6 [1.9] ha, median = 11.8 ha). CP1 whole fields ranged in size from 7.3 – 23.9 ha
(mean [SE] = 12.3 [2.1] ha, median = 10.9 ha), and the introduced portion of CP1 fields
ranged in size from 5.5 – 17.0 ha (mean [SE] = 9.4 [1.5], median = 8.0 ha).
Vegetation composition and structure were investigated using one randomly
located sampling point per acre (i.e., 1 sampling point per 0.4 ha) in each study field (n

20

=17 to 59 sampling points per field). Vegetation data were collected once in 2005 during
mid-summer (15 June – 14 July) and three times in 2006 during early summer (15 May –
14 June), mid-summer (15 June – 14 July), and late summer (15 July – 14 August).
Canopy height and litter depth were measured at each sampling point using a yardstick.
A modified 50 cm x 25 cm Daubenmire frame was centered on each sampling point
(Daubenmire 1959). All plant species within the Daubenmire frame were identified. The
percentage of Daubenmire frame area composed of bare ground, litter cover, and canopy
cover were visually estimated. Percent standing dead and living vegetation were also
visually estimated, and within the living canopy vegetation, the percent cover of grasses,
forbs, and woody vegetation were also determined. The total area of the Daubenmire
frame was accounted for as:

100% plot area = % bare ground + % litter cover + % total canopy cover

where:
% total canopy cover = % dead canopy vegetation + % live canopy vegetation
and:
% live canopy vegetation = % grasses + % forbs + % woody plants.

Boxplots were used to compare vegetation variables of CP23 native and CP1
introduced plant communities. Native plant communities were present in portions of the
CP1 fields in addition to occurring throughout the CP23 fields (Figure 2.1). CP1 native
grassland plantings ranged in size from 0.3–6.9 ha (mean [SE] = 4.0 [1.0], median = 4.2)
To investigate whether these two native plant communities provided similar potential
grassland bird habitat, the structure of CP23 native vegetation and the native portions of
21

CP1 fields were also compared using boxplots. Boxplot whiskers extended to the
minimum and maximum observed values unless outliers were present, in which case
whiskers extend 1.5 times the interquartile range and outliers are denoted outside the
range of the whiskers.
Avian data were collected using line-transect sampling (Edwards et al. 1981).
The total area of each study field (i.e., except CP23 wetlands and associated wetland
buffers) was surveyed every 2 weeks from 1 June – 14 August in 2005 (n=5 surveys in all
fields) and from 15 May – 14 August in 2006 (n=6 surveys per field). Surveys took place
along transects running the length of each field. The first transect was established 25 m
from a randomly selected field corner and subsequent transects were established at 50 m
intervals across the width of the field. An investigator recorded the species and location
of each individual bird seen or heard. If an individual bird was flushed by the observer,
the new location of the bird was noted in order to prevent duplicate counting of
individuals. If a bird was observed but could not be identified, the location of the bird
was recorded and the species designated as unknown. Unidentified birds were not
included in data analyses. Avian surveys were conducted after sunrise and before 12:00
p.m., and surveys were not performed in the rain or if wind speeds were over 16 kph.
Comparisons of avian density were completed at administrative (i.e., between
CP23 and CP1 fields) and ecological (i.e., between CP23 fields and the introduced
portion of CP1 fields) levels using boxplots with whisker lengths extending to the
minimum and maximum observed values unless outliers were present, in which case
whiskers extend 1.5 times the interquartile range and outliers are denoted outside the
range of the whiskers. Data from the introduced portion of all 7 CP1 fields were used

22

during ecological-level comparisons, but for administrative-level comparisons, data were
used from 5 of the 7 CP1 study fields because the native plant community portion of 2
CP1 fields did not establish successfully. For each grassland bird species, mean avian
density values in CP23 fields, CP1 fields, and the introduced portion of CP1 fields were
determined by averaging the density values of individual study areas of each field/cover
type. Avian species were categorized as reliably detected (i.e., detection probability
≥0.10) or unreliably detected (i.e., detection probability could not be estimated or
detection probability <0.10) in each CREP field and grassland type. Differences in bird
densities were evaluated only for those species reliably detected. Density was selected as
the measure of abundance for comparison in order to control for variation in study field
size and help account for passive sampling, which is that large habitat patches are more
likely to be occupied by a species than small patches through chance alone (Ribic et al.
2009).
Studies of species occurrence and abundance that assume perfect detection are
often biased by imperfect detection (i.e., a species was actually present but was not
observed) (MacKenzie et al. 2009). To account for imperfect detection, I estimated the
detection probability, p, of each observed species, and omitted species with low detection
probability estimates (i.e., p ≤ 0.10) from further analyses. Avian detection modeling
was conducted for species that were observed in at least 10% of the total number of
surveys (MacKenzie et al. 2002). Species that were observed in > 10% of surveys were
not included in density comparisons.
Detection probability of each avian species was estimated using single-season
models (MacKenzie et al.2002) in the program PRESENCE (Hines 2006). The use of

23

single season models was based on the results of likelihood ratio testing (MacKenzie et
al. 2006, Nichols et al. 2008), which indicated that the more highly parameterized multiseason models did not fit the data significantly better than single season models (p >
0.05).
For studies with small sample sizes, model ranking using Akaike‘s information
criterion corrected for small sample sizes (AICc) is recommended (Burnham and
Anderson 2002). The top-ranking detection model for each avian species was identified
as having the lowest AICc value among all candidate models with constant occupancy,
ψ(·) (MacKenzie et al. 2002). Candidate (i.e., successful) detection models were those
that achieved 1) numerical convergence of parameter estimates to >6 significant figures,
2) a balanced variance-covariance matrix, and 3) small- or moderately-sized standard
errors of betas, the untransformed estimates of coefficients for covariates (i.e., standard
error was no greater than 5 times the corresponding beta) (D. MacKenzie, Program
PRESENCE – FAQ at http://www.phidot.org/forum, accessed October 2009).
Detection probability estimates were obtained from the top-ranking detection
model for each species. For CREP administrative-scale analyses, potential detection
covariates were ordinal date and field type (i.e., CP23 or CP1). For ecological-level
analyses, potential detection covariates were ordinal date and plant community type (i.e.,
native or introduced). Ordinal date was standardized to reduce biases in parameter
estimation (MacKenzie et al. 2002). When plant community or field type and/or ordinal
date were covariates in the top-ranking detection model, multiple estimates of detection
resulted, and the lowest estimate was reported. For species where no detection models
were successfully generated using the full data set, detection probability estimates for
24

individual study sites were examined, and if detection probability estimates for a subset
of the data appeared adequate (e.g., detection estimates for all fields in 2006 were ≥0.10),
then detection modeling was performed on this data subset. For species where the naïve
estimate of occupancy, ψ, equaled 1, the fields were fully occupied and detection
modeling was conducted with ψ fixed to 1.
RESULTS
Vegetation results
Sixty-nine plant species were observed among all fields in this study. Sixty
species were found in CP23 native plant communities, and 37 species occurred in CP1
introduced plant communities (Table 2.2). Thirty-two plant species were unique to native
communities, and nine species occurred only in introduced communities (Table 2.2). The
most commonly observed grass species of native fields (i.e., observed in ≥ 20% of
vegetation sample plots) were Big Bluestem (Andropogon gerardii), Indiangrass
(Sorghastrum nutans), and Little Bluestem (Andropogon scoparius), while the most
frequently observed grass species in introduced plant communities were Orchardgrass
(Dactlyis glomerata), Timothy (Phleum pratense), and Annual Ryegrass (Lolium
perenne). The most commonly observed forbs in native fields included Dandelion
species (Taraxacum spp.), Thistle species (Cirsium spp.), Black Medick (Medicago
lupulina), Goldenrod species (Solidago spp.), Queen Anne‘s Lace (Daucus carota), and
Lettuce species (Lactuca spp.). The most frequently observed forb species in introduced
plant communities were Dandelion and Thistle species, along with Alfalfa (Medicago
sativa), an intentionally planted species.

25

Structural variables of the CP23 native plant communities and the introduced
portion of CP1 fields were summarized and compared for 1) 2005 mid-summer, 2) 2006
early summer, 3) 2006 mid-summer, and 4) 2006 late summer (Figures 2.2-2.5).
Bare ground tended to compose the smallest proportion of total sample plot area
(i.e., total plot area consists of bare ground, litter cover, and canopy cover) in native and
introduced plant communities; in native vegetation, bare ground made up the smallest
percentage of plot area throughout the avian breeding season in 2006, and in introduced
vegetation, bare ground made up the smallest proportion of plot area in 2005 and
throughout 2006 (Figures 2.2-2.5). In CP23 fields in 2006, bare ground ranged from 1.2
– 12.2% in early summer (mean = 7.5, SE = 1.4), from 0.8 – 14.3% in mid-summer
(mean = 6.5, SE = 1.6), and from 2.4 – 16.5% in late summer (mean = 7.5, SE = 1.7). In
introduced vegetation of CP1 fields in 2005, bare ground ranged from 0 – 6.3% in midsummer (mean = 2.5, SE = 1.0), and in 2006, bare ground ranged from 0 – 10.6% in early
summer (mean = 3.2, SE = 1.4), from 0 – 3.8% in mid-summer (mean = 1.1, SE = 0.6),
and from 0 – 3.8% in late summer (mean = 1.1, SE = 0.5).
The total plant canopy (i.e., living and standing dead vegetation) tended to
compose the largest proportion of sample plot area in native and introduced plant
communities; in native vegetation, the total plant canopy made up the largest proportion
of sample plot area in mid-summer 2005 and in mid- and late summer 2006, and in
introduced vegetation, the total plant canopy composed the largest proportion of plot area
in mid-summer 2005 and throughout 2006 (Figures 2.2-2.5). In CP23 fields in 2005,
total canopy cover ranged from 54.8 – 72.1% in mid-summer (mean = 66.1, SE = 3.2),
and in 2006, total canopy cover ranged from 36.3 – 71.1% in mid-summer (mean = 58.3,

26

SE = 5.6) and from 49.1 – 73.7% in late summer (mean = 59.4, SE = 3.3). In introduced
vegetation of CP1 fields in 2005, total canopy cover ranged from 71.3 – 95.6% in midsummer (mean = 87.9, SE = 3.8), and in 2006, total canopy cover ranged from 28.7 –
85.3% in early summer (mean = 60.3, SE = 8.1), from 45.4 – 84.4% in mid-summer
(mean = 63.8, SE = 5.4), and from 47.0 – 91.1% in late summer (mean = 69.4, SE = 6.5).
Standing dead vegetation made up a small portion of the total canopy within
native and introduced plant communities throughout the study (Figures 2.2-2.5). In CP23
fields in 2005, dead vegetation in the canopy ranged from 0.7 – 4.6% (mean = 2.4, SE =
0.6) in mid-summer, and in 2006, standing dead vegetation ranged from 1.7 – 6.9% in
early summer (mean = 42.5, SE = 7.6), from 0.1 – 6.6% in mid-summer (mean = 2.5, SE
= 0.9), and from 0.1 – 3.4% in late summer (mean = 1.5, SE = 0.5). In introduced
vegetation of CP1 fields in 2005, standing dead vegetation ranged from 0 – 11.8% in
mid-summer (mean = 2.3, SE = 1.6), and in 2006, standing dead vegetation ranged from
0 – 2.6% in early summer (mean = 1.0, SE = 0.4), from 0 – 0.8% in mid-summer (mean =
0.3, SE = 0.1), and from 0 – 5.1% in late summer (mean = 0.9, SE = 0.7).
There were virtually no woody plants in the live canopy of both native and
introduced vegetation, in both years and all time periods of the study (Figures 2.2-2.5).
With the absence of woody vegetation, the live canopy was composed only of forbs and
grasses, thus the relationship of grasses to forbs in the living canopy was inversely
proportional, i.e., if the percentage of grasses in the living canopy increased, the
percentage of forbs decreased. As expected for the warm-season grasses of native
plantings, grasses tended to be less abundant than forbs in the live canopy of CP23 fields
in early summer 2006 (Figure 2.2; forbs: range 8.3 – 41.1%, mean = 27.5, SE = 5.9;

27

grasses: range 6.4 – 33.3%, mean = 15.0, SE = 3.3), but by mid-summer forbs and
grasses were equally abundant in native vegetation (Figure 2.4; forbs: range 9.0 – 52.1%,
mean = 29.0, SE = 5.8; grasses: 21.0 – 37.6%, mean = 26.5, SE = 2.1), and by late
summer grasses were more abundant than forbs (Figure 2.5; forbs: range 9.6 – 41.0%,
mean = 22.2, SE = 4.6; grasses: 24.3 – 54.6%, mean = 35.3, SE = 4.1). Within
introduced vegetation, forbs and the cool-season grasses were equally abundant in the
live canopy throughout the summer in 2006 (Figures 2.3-2.5). In early summer 2006,
forbs in the living canopy of introduced vegetation ranged from 8.7 – 49.8% (mean =
29.0, SE = 5.7) and grasses ranged from 19.3 – 42.4% (mean = 30.3, SE = 3.4); in midsummer forbs ranged from 14.9 – 41.5% (mean = 36.4, SE = 6.1) and grasses ranged
from 22.3 – 34.1% (mean = 27.0, SE = 1.5); and in late summer forbs ranged from 20.8 –
61.4% (mean = 37.7, SE = 6.0) and grasses ranged from 23.3 – 37.2% (mean = 30.8, SE
= 2.4).
In addition to different patterns of change in the ratio of grasses and forbs within
native and introduced plant canopies during the course of the avian breeding season,
other differences in the structure of native and introduced vegetation existed. In
particular, native plant communities tended to have more bare ground and less total plant
canopy cover than introduced plant communities (Figures 2.2-2.5). Less bare ground was
present in introduced vegetation than in native fields throughout the study, and more
canopy cover existed in introduced than native vegetation in mid-summer 2005 and early
and mid-summer 2006 (Figures 2.2-2.5). Bare ground in CP23 vegetation in 2005 ranged
from 9.5 – 24.5% (mean = 15.9, SE = 2.4); total canopy cover in native vegetation in
early summer 2006 ranged from 22.9 – 64.5% (mean = 46.2, SE = 7.1); see bare ground

28

and total canopy cover results above for other range and mean information. Differences
between native and introduced plant communities may be most important for grassland
birds during early summer when breeding territories are selected. In early summer 2006,
native vegetation had a larger proportion of litter cover than introduced vegetation
(Figure 2.3). Litter cover in early summer 2006 ranged from 25.6 – 76.0% (mean = 46.3,
SE = 7.9) in native vegetation and from 14.7 – 70.1% (mean = 36.6, SE = 7.3) in
introduced vegetation. Also, since native vegetation tends to remain upright during
winter whereas introduced grasses generally collapse (USDA 2000a, 2000b), it was not
surprising that native plant communities had more standing dead vegetation in the canopy
than introduced vegetation in early summer 2006 (Figure 2.3; see standing dead results
above for range and mean information).
Native plant communities were present in portions of the CP1 fields as well as
throughout CP23 fields. To examine whether these two types of native plant
communities provided similar potential grassland bird habitat, the composition and
structure of CP23 native vegetation and the native portions of CP1 fields were compared.
The composition of CP23 and CP1 native vegetation was similar, with few
differences in the most commonly observed plant species (i.e., observed in ≥ 20% of
vegetation sample plots). In CP1 native vegetation the most common plant species were
Big Bluestem, Indiangrass, Kentucky Bluegrass (Poa pratensis), Dandelion species,
Thistle species, Goldenrod species, Queen Anne‘s Lace, and Shasta Daisy
(Chrysanthemum x superbum ‘Alaska’), and the most common plant species in CP23
fields were Big Bluestem, Indiangrass, Little Bluestem, Dandelion species, Thistle
species, Goldenrod species, Queen Anne‘s Lace, Black Medick, and Lettuce species.

29

Structural variables of CP23 native vegetation and the native portions of CP1
fields were summarized and compared for 1) 2005 mid-summer; 3) 2006 early summer;
4) 2006 mid-summer; and 5) 2006 late summer (Figures 2.6-2.9).
Structural characteristics of both types of native plantings were similar (Figures
2.6-2.9); only a few differences in the vegetation of CP23 and CP1 native plant
communities were observed. In late summer 2006, CP23 native vegetation had less total
canopy cover and slightly more bare ground than CP1 native vegetation (Figure 2.9). In
late summer 2006, bare ground ranged from 2.4 – 16.5% (mean = 7.5, SE = 1.7) in CP23
fields and from 1.1 – 7.9% (mean = 4.1, SE = 1.4) in CP1 native vegetation; total canopy
cover ranged from 49.1 – 73.7% (mean = 59.4, SE = 3.3) in CP23 fields and from 70.9 –
77.3% (mean 71.1, SE= 3.1) in CP1 native plant communities. Also, in mid-summer
2005, more standing dead vegetation was present in CP23 fields than in CP1 native plant
communities (Figure 2.6), but the opposite was true the following year, when in midsummer 2006, CP23 fields contained slightly less standing dead vegetation than CP1
native plant communities (Figure 2.8). In mid-summer 2005, standing dead vegetation
ranged from 0.7 – 4.6% (mean = 2.4, SE = 0.6) in CP23 fields and from 0.2 – 0.9%
(mean = 0.5, SE = 0.1) in CP1 native vegetation, and in mid-summer 2006, standing dead
vegetation ranged from 0.1 – 6.6% (mean = 2.5, SE = 0.9) in CP23 fields and from 1.8 –
8.4% (mean = 4.4, SE = 1.4) in CP1 native plant communities.
Avian results
A total of 34 avian species were observed in CREP study fields during the 2 years
of this study (Table 2.3). In 2005, 23% of individual birds observed during surveys were
unidentified, and in 2006 surveyors were more experienced and only 8% of observed

30

individuals were unidentified. Using the grassland bird list developed for the BBS by the
U.S. Geographic Survey (Sauer et al. 1995), 11 of the observed species were identified as
grassland specialists: Bobolink (Dolichonyx oryzivorus), Dickcissel (Spiza americana),
Eastern Meadowlark (Sturnella magna), Grasshopper Sparrow (Ammodramus
savannarum), Henslow‘s Sparrow (Ammodramus henslowii), Horned Lark (Eremophila
alpestris), Northern Harrier (Circus cyaneus), Ring-necked Pheasant, Savannah Sparrow
(Passerculus sandwichensis), Sedge Wren (Cistothorus platensis), and Vesper Sparrow
(Pooecetes gramineus). Six of these grassland specialist species are of special
conservation priority in Michigan (Table 2.3). Bobolink, Dickcissel, Grasshopper
Sparrow, Henslow‘s Sparrow, Northern Harrier, and Sedge Wren are designated as
Michigan priority landbirds by the bird conservation group Partners in Flight (Knutson et
al. 2001, Matteson et al. 2009), while Dickcissel, Grasshopper Sparrow, and Northern
Harrier are also state species of special concern and Henslow‘s Sparrow is a state
endangered species (Table 2.3).
Avian detection probability modeling
Three grassland bird species were too rarely observed to estimate their detection
probabilities (i.e., observed in less than 10% of surveys): Henslow‘s Sparrow, Horned
Lark, and Northern Harrier. Detection probability modeling was performed for the 8
other grassland specialist species observed during this study: Bobolink, Dickcissel,
Eastern Meadowlark, Grasshopper Sparrow, Ring-necked Pheasant, Savannah Sparrow,
Sedge Wren, and Vesper Sparrow. Likelihood ratio testing indicated that more highly
parameterized multi-season models did not fit the data significantly better than simpler

31

single-season models (p > 0.05), so single season models were used in all detection and
occupancy modeling efforts (MacKenzie et al. 2006, Nichols et al. 2008).
At the administrative level of analysis (i.e., CP23 and CP1 fields), all 8 grassland
species were reliably detected. Adequate detection probability estimates (i.e., p ≥ 0.10)
were achieved for all 8 species, with detection probability estimates ranging from 0.12 to
0.81 (Table 2.4). However, an adequate detection probability estimate was obtained only
when a subset of the data was used in the case of one species, Sedge Wren (2006 data),
most likely because Sedge Wren was rare in 2005 (mean density = 0.01, SE = 0.01 in
CP23 fields; mean density = 0.0, SE = 0.0 in introduced vegetation of CP1 fields). CREP
field type and/or standardized ordinal date were included in the best detection models for
Bobolink, Eastern Meadowlark, Ring-necked Pheasant, Savannah Sparrow, and Sedge
Wren (Table 2.4). The detection probability of Bobolink varied between 2005 and 2006,
but detection probability estimates were >0.63 regardless of year, indicating that
Bobolink was readily detected throughout each breeding season (Figure 2.10). Eastern
Meadowlark was reliably detected in CP23 and CP1 fields, but the species tended to be
more likely to be detected in CP23 fields, although the 95% confidence intervals for
detection estimates of the two field types were overlapping (i.e., 95% confidence interval
was 0.10 – 0.16 in CP23 fields and 0.08 – 0.15 in CP1 fields; Figure 2.11). Ring-necked
Pheasant detection probability varied between 2005 and 2006, but detection probability
estimates were >0.17 in both years, indicating that the species was reliably detected
throughout both breeding seasons (Figure 2.12). Ring-necked Pheasant was adequately
detected in CP23 and CP1 fields, but the species tended to be more likely to be detected
in CP23 fields, although the 95% confidence intervals for detection estimates of the two

32

field types were overlapping (i.e., 95% confidence interval was 0.21 – 0.46 in CP23
fields and 0.05 – 0.32 in CP1 fields; Figure 3.4). Detection probability estimates for
Savannah Sparrow were >0.81 regardless of year, indicating that the species was reliably
detected throughout each breeding season (Figure 2.13). The detection probability of
Sedge Wren was >0.15 throughout 2006, indicating that the species was reliably detected
throughout the breeding season, but Sedge Wren detection probability was positively
associated with standardized ordinal date, suggesting that optimal detection of this
species occurred toward the end of the breeding season (Figure 2.14).
At the ecological level of analysis (i.e., native and introduced grasslands), all 8
grassland species were reliably detected. Adequate detection probability estimates (i.e., p
≥ 0.10) were obtained for all 8 grassland species, with detection probability estimates
ranging from 0.12 to 0.84 (Table 2.4). Standardized ordinal date was included in the best
detection model for Dickcissel (Table 2.4). The detection probability for Dickcissel
varied between 2005 and 2006, but detection probability estimates were > 0.12 during
both years, indicating that Dickcissel was reliably detected throughout both breeding
seasons (Figure 2.15) The best detection model for Sedge Wren included grassland type,
and the species tended to be more likely to be detected in native grasslands (p = 0.83 [SE
= 0.16] ) than in introduced grasslands (p = 0.14 [SE = 0.16]), but was reliably detected
in both cover types (Figure 2.16). For the remaining 6 grassland bird species the best
detection model included only the intercept (Table 2.4).
Avian density comparisons
Avian species were categorized as reliably detected (i.e., p ≥0.10) or unreliably
detected (i.e., p could not be estimated or p <0.10) in each CREP field/grassland type.

33

Avian densities were reported (Table 2.3) and compared (Figures 2.17-2.24) only for
those species that were reliably detected. Species that were not reliably detected were
designated as present if observed in study fields/grasslands and as undetected if not
observed (i.e., species that were not observed in fields/grasslands could not be assumed to
be absent if detection probability estimates were <0.10; Table 2.3).
Ten grassland bird species were observed in 2005; 9 grassland species were found
in native grasslands and 7 species occurred in the introduced portion of CP1 fields and in
CP1 whole fields (Table 2.3). Of the 8 grassland bird species reliably detected (i.e., p ≥
0.10) in 2005, Grasshopper Sparrow and Sedge Wren were unique to native vegetation
and no species were unique to introduced plant communities. Note that in 2005,
Northern Harrier, was detected in introduced vegetation of CP1 fields but is recorded as
undetected for CP1 whole fields because data were used from only 5 of the 7 CP1 study
fields (i.e., data from fields where the native plant community portion did not establish
successfully were omitted (Table 2.3)). In 2006, 11 grassland bird species were observed
(Table 2.3); all 11 species were found in native plant communities and 8 species were
seen in the introduced portion of CP1 fields and in CP1 whole fields (Table 2.3). All 8
grassland species that were reliably detected (i.e., p ≥ 0.10) in 2006 occurred in all
field/grassland types (Table 2.3).
Among adequately detected species (i.e., p ≥ 0.10), the most abundant grassland
birds were Savannah Sparrow and Bobolink (Table 2.3, Figures 2.17-2.24). In 2005,
Savannah Sparrow density ranged from 0.14 – 3.14 individuals/ha in CP23 fields (median
= 0.58, mean = 0.98, SE = 0.39), from 0.03 – 0.79 individuals/ha in introduced vegetation
(median = 0.29, mean = 0.37, SE = 0.12), and from 0.18 – 0.95 individuals/ha in CP1

34

fields (median = 0.51, mean = 0.55, SE = 0.17; Table 2.3, Figure 2.17). In 2006,
Savannah Sparrow density ranged from 0.55 – 5.16 individuals/ha in CP23 fields (median
= 1.64, mean = 2.14, SE = 0.62), from 0.12 – 1.15 individuals/ha in introduced vegetation
(median = 0.83, mean 0.94, SE = 0.32), and from 0.06 – 2.81 individuals/ha in CP1 fields
(median = 1.13, mean = 1.22, SE = 0.46; Table 2.3, Figure 2.17). In 2005, Bobolink
density ranged from 0 – 0.94 individuals/ha in CP23 fields (median = 0.34, mean = 0.38,
SE =0.15), from 0.09 – 2.55 individuals/ha in introduced vegetation (median =1.44, mean
= 1.33, SE = 0.37), and from 1.01 – 5.54 individuals/ha in CP1 fields (median = 1.46,
mean = 2.28, SE = 0.93; Table 2.3, Figure 2.18). In 2006, Bobolink density ranged from
0 – 1.44 individuals/ha in CP23 fields (median = 0.18, mean = 0.60, SE = 0.26), from 0 –
1.71 individuals/ha in introduced vegetation (median = 0.44, mean 0.68, SE = 0.27), and
from 0 – 0.80 individuals/ha in CP1 fields (median = 0.30, mean = 0.35, SE = 0.15; Table
2.3, Figure 2.18).
One influence on grassland bird density in each study field was the breeding
territory size of each species, which is affected by a variety of factors, including habitat
suitability (e.g., Lanyon 1995) and local population density (e.g., Harmeson 1974 cited
from Temple 2002). For each grassland bird species, the territory size range in Michigan
was estimated using information from species accounts in the Cornell Lab of
Ornithology‘s Birds of North America Online (BNA Online) (Lanyon 1995, Martin et al.
1995, Vickery 1996, Giudice and Ratti 2001, Herkert et al. 2001, Jones and Cornely
2002, Temple 2002, Wheelwright and Rising 2008; Table 2.5). Based on the size ranges
of 1) estimated breeding territories for each grassland bird species in Michigan and 2)
CREP study fields, CP23 grasslands could support from 1 – 180 grassland bird breeding

35

territories (e.g., CP23 native study fields provided from 5 – 23 breeding territories for
Grasshopper Sparrow), while CP1 whole fields could provide from 1 – 218 grassland bird
breeding territories, and CP1 introduced grasslands could support from 1 – 155 grassland
bird breeding territories (Table 2.5). Overall, CP23 fields provided 0- 22% fewer
breeding territories than CP1 whole fields (Table 2.5). However, CP23 fields supported 0
– 25% more breeding territories than CP1 introduced grasslands (e.g., CP23 fields
provided 0 – 14% more breeding territories for Eastern Meadowlark than did CP1
introduced grasslands) (Table 2.5).
Avian density comparisons were conducted for species that were reliably detected
(i.e., p ≥ 0.10): Bobolink, Dickcissel, Eastern Meadowlark, Grasshopper Sparrow, Ringnecked Pheasant, Savannah Sparrow, Sedge Wren, and Vesper Sparrow. Boxplots were
used to compare avian densities in 1) CP23 and CP1 fields and in 2) CP23 fields and the
introduced vegetation of CP1 fields (Figures 2.17-2.24).
In 2005, Bobolink density was lower in CP23 fields than in CP1 introduced
vegetation or in CP1 fields, and no differences in Bobolink density were observed in
2006 (see above for range, median, and mean information; Figure 2.18). No differences
between Dickcissel density in CP23 native fields and CP1 introduced vegetation or in
CP23 and CP1 fields were observed in 2005 or 2006 (Figure 2.19). Eastern Meadowlark
density also did not differ among field or grassland types during both years of the study
(Figure 2.20). In contrast, Grasshopper Sparrow density was higher in CP23 fields than
in introduced vegetation or in CP1 fields in 2005 and 2006 (Figure 2.21). In 2005,
Grasshopper Sparrow density ranged from 0.01 – 0.14 individuals/ha in CP23 fields
(median = 0.04, mean = 0.05, SE =0.02, and was 0 in CP1 native vegetation and CP1

36

whole fields (Table 2.3, Figure 2.21). In 2006, Grasshopper Sparrow density ranged
from 0 – 0.16 individuals/ha in CP23 fields (median = 0.08, mean = 0.08, SE = 0.03),
from 0 – 0.03 individuals/ha in introduced vegetation (median = 0, mean 0.01, SE =
0.01), and from 0 – 0.02 individuals/ha in CP1 fields (median = 0, mean = 0.01, SE =
0.01; Table 2.3, Figure 2.21). Ring-necked Pheasant density was higher in CP23 fields
than in introduced vegetation in 2005, and no differences in Ring-necked Pheasant
density were observed in 2006 (Figure 2.22). In 2005, Ring-necked Pheasant density
ranged from 0.02 – 0.73 individuals/ha in CP23 fields (median = 0.06, mean = 0.15, SE
=0.10) and from 0 – 0.12 individuals/ha in introduced vegetation (median =0.01, mean =
0.03, SE = 0.02; Table 2.3, Figure 2.22). In 2005 and 2006, Savannah Sparrow density
was higher in CP23 fields than in introduced vegetation, and in 2006, Savannah Sparrow
density was also higher in CP23 fields than in CP1 fields (see above for range, median,
and mean information; Figure 2.17). Sedge Wren density was slightly higher in CP23
fields than CP1 fields in 2006 (Figure 2.23). In 2006, Sedge Wren density ranged from 0
– 0.14 individuals/ha in CP23 fields (median = 0.06, mean = 0.07, SE = 0.02) and from 0
– 0.18 individuals/ha in CP1 fields (median = 0, mean = 0.03, SE = 0.02; Table 2.3,
Figure 2.23). In 2005, Vesper Sparrow density was higher in CP23 fields than introduced
vegetation or in CP1 fields, and no differences in Vesper Sparrow density were observed
in 2006 (Figure 2.24). In 2005, Vesper Sparrow density ranged from 0 – 0.18
individuals/ha in CP23 fields (median = 0.06, mean = 0.07, SE =0.03), from 0 – 0.04
individuals/ha in introduced vegetation (median = 0, mean 0.01, SE = 0.01), and from 0 –
0.04 individuals/ha in CP1 fields (median = 0.01, mean = 0.02, SE = 0.01; Table 2.3,
Figure 2.24).

37

DISCUSSION
Vegetation sampling of CP23 fields and the introduced portion of CP1 fields
resulted in the observation of 73 plant species, 32 of which were unique to CP23 plant
communities, including all native grass species and a majority of native forbs. Plant
species richness was greatest in CP23 fields, which contained over 1.5 times more plant
species than the introduced vegetation of CP1 fields (i.e., 60 spp. vs. 37 spp.). During
the avian breeding season, native plant communities tended to have more bare ground
and less total plant canopy cover than introduced vegetation. Native and introduced plant
communities also exhibited different patterns of change in the ratio of grasses and forbs
within the plant canopy throughout the avian breeding season. Warm-season native
grasses were less abundant than forbs in CP23 fields in early summer, but by midsummer grasses and forbs were equally abundant, and by late summer grasses were more
abundant than forbs. Cool-season introduced grasses were equally abundant in CP1
introduced plant communities throughout the summer. Differences between native and
introduced plant communities may be most important for grassland birds during breeding
territory selection, and early summer differences in the vegetation structure of CP23
native and CP1 introduced vegetation included a larger proportion of litter cover and
more standing dead vegetation in CP23 native grasslands. The larger amount of standing
dead vegetation in CP23 fields in early summer indicates that the native plant
communities in Michigan CREP grasslands are providing more desirable winter habitat
for overwintering wildlife such as Ring-necked Pheasant than are the introduced portions
of CP1 fields (USDA 2000a, 2000b).

38

Avian surveys of CP23 and CP1 CREP grasslands in Michigan‘s Saginaw Bay
watershed resulted in the observation of 34 species, including 11 grassland associates
(Sauer et al. 1995). Survey results indicate that CREP grasslands provide habitat for a
majority of grassland bird species found in Michigan (i.e., 11 of 15 species documented
to currently occur in Michigan; BBAE 2010). Grassland bird species observed in CREP
study fields included all 4 species found to have significantly declining population trends
(p < 0.10) in Michigan during the most recent analysis of Breeding Bird Survey (BBS)
data (i.e., Bobolink, Eastern Meadowlark, Ring-necked Pheasant; Sauer et al. 2008), as
well as 1 state endangered species (i.e., Henslow‘s Sparrow), and 2 species of state
special concern (i.e., Dickcissel and Grasshopper Sparrow), suggesting that CREP
grasslands provide habitat for the most imperiled grassland species in Michigan.
Eight grassland bird species were detected frequently enough (i.e., in ≥10% of
surveys) to allow detection probability modeling: Bobolink, Dickcissel, Eastern
Meadowlark, Grasshopper Sparrow, Ring-necked Pheasant, Savannah Sparrow, Sedge
Wren, and Vesper Sparrow. Detection estimates were adequate for all 8 grassland bird
species and were generally high (i.e., p > 0.25 for 6 species in CP1 and CP23 fields
and/or CP23 fields and CP1 introduced vegetation), indicating that the survey technique
used was an effective methodology for documenting grassland bird occurrence in the
study system, and providing a sound basis for reliable density comparisons.
For 5 out of 8 grassland bird species included in density comparisons, higher
densities were associated with CP23 fields than with CP1 whole fields and/or the
introduced portion of CP1 fields during at least one year of the study (i.e., Grasshopper
Sparrow, Ring-necked Pheasant, Savannah Sparrow, Sedge Wren, and Vesper Sparrow).

39

The densities of these 5 grassland bird species were higher in CP23 fields than in CP1
whole fields even though CP1 fields could support 0 - 18% more grassland bird breeding
territories for these species than CP23 fields, suggesting that CP1 fields containing
introduced and native grasslands are not as attractive to several breeding grassland bird
species as similar-sized or smaller CP23 fields containing only native grasslands. The
densities of 2 species, Grasshopper Sparrow and Savannah Sparrow, showed particularly
consistent trends, with higher densities in CP23 fields than in the introduced portion of
CP1 fields and/or CP1whole fields during both years of the study. However, one species
of management priority, Bobolink, was more abundant in the introduced vegetation of
CP1 fields during the first year of the study, demonstrating the diverse response of
grassland birds to habitat features. This observation provides support for using speciesspecific habitat preference information when managing grassland bird species (e.g.,
Murray et al. 2008).
I attempted to quantify the productivity of grassland bird species found in the
native and introduced CREP grasslands of this study (Appendix 2.3). However, the
intensive nest survey techniques used were unsuccessful in my study system (i.e., a total
of 3 active grassland bird nests were located in all fourteen 16.8 ha survey areas), largely
due to the highly cryptic nature of grassland bird nests (Appendix 2.3). For example,
even when a nest was assumed to be in a particular area due to adult bird behaviors such
as food transport, nests were generally impossible to locate below the grass canopy
without severely degrading the vegetation structure and litter layer of the nest vicinity.
There is little doubt that density information is most informative about habitat quality
when coupled with information about reproduction and survival of the species of interest.

40

Van Horne (1983) raised consciousness that habitat quality as a measure of the
importance of a habitat in maintaining the population of a species ―should be defined in
terms of the survival and production characteristics, as well as the density, of the species
occupying that habitat.‖ However, recently other researchers have provided evidence
that density alone can be a reliable indicator of habitat quality (Bock and Jones 2004,
Perot and Villard 2009). Bock and Jones (2004) surveyed 109 published studies of 67
European or North American bird species and found that habitats with higher densities
displayed greater avian reproduction rates (i.e., greater recruitment per capita and per unit
of land area) in the majority of cases (i.e., ≥72%. Bock and Jones (2004) concluded that
―in most cases, density will be a reliable indicator of habitat quality, and bird-count data
will be an appropriate basis from which to make land-management and conservation
decisions.‖
Other researchers have also found differences in grassland bird abundances
between native and introduced grasslands. In a comparison of avian abundance, richness,
and productivity on privately owned warm- and cool-season grass farm fields in
Pennsylvania, Giuliano and Daves (2002) found that warm-season grasslands supported a
greater abundance and richness of birds, including Grasshopper Sparrow, Vesper
Sparrow, and other species of conservation priority.
In contrast, other studies in similar systems found no differences between
grassland bird abundances in native and introduced vegetation or found greater grassland
bird abundances in introduced vegetation (e.g., McCoy et al. 2001, Wentworth 2010). In
these studies, the structure and composition of native grasslands differed, often widely,
from the native fields I examined, tending to be less diverse, have a denser canopy,

41

and/or to be poorly established or switchgrass (Panicum virgatum) dominated (i.e., dense
stands of switchgrass have been found to be poor grassland bird habitat; Norment et al.
1999). Wentworth et al. (2010) demonstrated that in Pennsylvania CREP fields,
grassland obligate species were rare but were found in greater densities on fields of coolseason grasses than in fields of CP2 or mixed grasses. However, CP2 fields in the study
were not yet completely established and some were actually dominated by cool-season
grasses or other vegetation types (Wentworth et al. 2010). In a study of Conservation
Reserve Program (CRP) grasslands in Missouri, McCoy et al. (2001) found that more
grassland bird species occurred at higher abundances in introduced/CP1 fields than in
native/CP2 fields, which were monocultures of switchgrass and were less diverse and
denser than the CP1 fields in their study. The opposite was true of CP23 CREP fields in
my study; CP23 fields were more diverse and less dense (i.e., with sparser canopy cover
and more visible bare ground) than introduced vegetation of CP1 fields.
Differences in the findings of this study and other investigations underscore the
importance of evaluating the effects of CREP and similar programs on grassland birds at
local scales (e.g., ecoregions, or major watersheds). It is important to note that the
consistency of results from studies in different regions may be affected by differences in
grassland composition, age or successional stage, and structure.
Management recommendations
The results of this study indicate that Michigan CREP grasslands, especially
native CP23 grasslands, provide grassland bird habitat, a critical resource for an avian
guild whose continued decline has been identified as an impending conservation crisis
(Knopf 1996, Brennan and Kuvlesky 2005). Density comparisons indicate that CP23

42

grassland plantings supported greater densities of more grassland bird species than the
introduced vegetation of CP1 fields or CP1 whole fields. Despite the greater cost of
native seed mixes (i.e., $16.00/pound v. $1.90/pound for introduced vegetation in 2005 in
Tuscola County, Michigan) and the specialized planting techniques required for
successful native vegetation establishment (e.g., use of native seed drills for planting the
fluffy seed of some native grass species), I recommend that CP23 whole-field native
plantings be utilized in favor of CP1 plantings within CREP in the Saginaw Bay area of
Michigan. One species of management priority, Bobolink, was more abundant in the
introduced vegetation of CP1 fields during the first year of this study. Because these
results demonstrate a diverse response of grassland birds to CREP grasslands, I further
recommend that the effect of similar grasslands be evaluated for individual species within
the grassland bird guild rather than for grassland birds as a group and that speciesspecific information be used in developing grassland bird conservation strategies.
Maintaining and/or increasing the area of CREP grasslands, especially native
grasslands, should be a grassland bird conservation strategy in Michigan. Threats to the
persistence and/or increase of farmland set-aside grasslands exist, and include
reconversion to corn or other crop. Recently, increased demand for biofuels has resulted
in a trend toward reconversion of agricultural set-aside lands into crop production (Secchi
and Babcock 2007, Searchinger et al. 2008, cited from Fargione et al. 2009), a trend that
is likely to continue due to large, federally mandated increases in biofuel production (i.e.,
>700% over 2006 production by 2022) (Fargione et al. 2009). Fargione et al. (2009)
suggested minimizing the negative effects of increased biofuel demand on wildlife by
utilizing wildlife-compatible biomass sources that do not require use of additional

43

production lands (e.g., wastes, cover crops, algae), and by maximizing the area of
perennial grasslands, including diverse native prairie plantings, to produce biofuel in
ways that are compatible with wildlife (e.g., harvesting biomass only after completion of
the avian breeding season). Landowner participation is required for the maintenance of
Michigan CREP lands. Other research has examined the response of landowners to land
retirement incentives and found a positive response to CREP farmland set-aside
incentives and that an increase in the incentive payment made at the time of enrollment
generates a greater increase in enrollment than a small raise in the annual payments made
over the term of the set-aside contract (Suter et al. 2008). I recommend that program
administrators consider using this method of promoting the retention and increase of
CREP lands in order to ensure the continued presence of grassland bird habitat the
program provides.

44

APPENDIX 2.1: CHAPTER 2 TABLES AND FIGURES

45

CP1 Field
(standard)
n=5

CP23 Field
n=7

Native
Vegetation

Example
Wetland and
25 m Wetland
Buffer

Introduced
Vegetation
(70% area)

CP1 Field
(native planting failed)
n=2
Native
Vegetation
(30% area)

Introduced
Vegetation

Fig. 2.1. Location, type, and number of Conservation Reserve Enhancement Program
(CREP) study fields within Tuscola County, Saginaw Bay watershed, Michigan, 2005
and 2006. Wetlands and wetland buffers were excluded from avian surveys and
vegetation sampling. The surveyed (i.e., upland) area of CP23 fields ranged in size from
6.9 – 19.8 ha (median [SE] = 12.6 [1.9] ha, median = 11.8 ha). The introduced portion of
CP1 fields ranged in size from 5.5 – 17.0 ha (mean [SE] = 9.4 [1.5], median = 8.0 ha),
and CP1 whole fields ranged in size from 7.3 – 23.9 ha (mean [SE] = 12.3 [2.1] ha,
median = 10.9 ha).

46

Table 2.1. Plant species and proportions of species included in the seed mixtures used to
establish the native and introduced vegetation communities of Conservation Reserve
Enhancement Program (CREP) grasslands of Michigan‘s Saginaw Bay watershed, 2002.
% Grass,
% Total Wildflower,
Seed
or Legume
Common Name
Scientific Name
Mixture
Mixture
CP23 Fields and Native Section of CP1 Fields - planted at 6.5 pounds/acre (16.1 #/ha)
Native Grasses
92.3
Big Bluestem*
Andropogon gerardii
33.3
Indiangrass*
Sorghastrum nutans
33.3
Little Bluestem*
Andropogon scoparius
33.3
Plateau® Tolerant Wildflower Mixture
7.7
Black Eyed Susan*
Rudbeckia hirta
10.0
Blanket Flower, Perennial
Gaillardia aristata
2.0
Catchfly
Silene armeria
2.0
Clover, Crimson
Trifolim incarnatum
10.0
Coneflower, Prairie, 'Mexican Hat'
Ratibida columnifera
3.0
Coneflower, Purple*
Echinacea purpurea
2.0
Coneflower, Yellow*
Ratibida columnifera
5.0
Coreopsis, Lance Leaf*
Coreopsis lanceolata
10.0
Coreopsis, Plains
Coreopsis tinctoria
3.0
Cosmos, ‗Sensation‘ Mix
Cosmos bipinnatus
5.0
Cosmos, Sulphur
Cosmos sulphureus
5.0
Daisy, Shasta, 'Alaska'
Chrysanthemum x superbum
4.0
Dame‘s Rocket
Hesperis matronalis
5.0
Flax, Perennial
Linum perenne lewisii
10.0
Lupine, Perennial*
Lupinus perennis
4.0
Poppy, California
Eschscholzia californica
5.0
Poppy, Corn, ‗Shirley Single‘ Mix Papaver rhoeas
5.0
Wallflower, Siberian
Erysiumum x marshallii
10.0
Introduced Section of CP1 Fields - planted at 11 pounds per acre (27.2 #/ha)
Introduced Grasses
45.5
Orchardgrass
Dactlyis glomerata
50.0
Timothy
Phleum pratense
50.0
Introduced Legumes
54.5
Alfalfa
Medicago sativa
50.0
Red Clover
Trifolium pratense
50.0
*Michigan native species (Voss 1985, Voss 1996, USDA NRCS 2011)

47

Table 2.2. Plant species observed in native and introduced plant communities of
Conservation Reserve Enhancement Program (CREP) grasslands, Saginaw Bay
watershed, Michigan, 2005 and 2006.
Plant Community Type
Common Name
Scientific Name
Native
Introduced
Alfalfa
Medicago sativa
n
i
Alsike Clover
Trifolium hybridum
n
i
Annual Ryegrass
Lolium perenne
n
i
Big Bluestem
Andropogon gerardii
n
Black Medick
Medicago lupulina
n
i
Black-eyed Susan
Rudbeckia hirta
n
i
Blanket Flower
Gaillardia aristata
n
Burdock
Arctium spp.
i
Canada Thistle
Cirsium arvense
n
i
Catchfly
Silene spp.
n
Cattail
Typha spp.
n
Chickweed
Stellaria spp.
n
Chicory
Cichorium intybus
i
Cinquefoil
Potentilla spp.
n
Common Mullein
Verbascum thapsus
n
Common Sow Thistle
Sonchus oleraceus
n
i
Corn Poppy
Papaver rhoeas
n
Cottonwood
Populus spp.
n
Curly Dock
Rumex crispus
n
Daisy Fleabane
Erigeron annuus
n
i
Daisy spp.
Asteraceae spp.
n
Dame's Rocket
Herperis matronalis
i
Dandelion
Taraxacum spp.
n
i
Downy Brome (Cheat Grass) Bromus tectorum
n
i
Fescue
Festuca spp.
n
i
Field Bindweed
Convolvulus arvensis
n
i
Fleabane
Erigeron spp.
n
Forked Catchfly
Silene dichotoma
n
Goat‘s-beard
Tragopogon spp.
n
Goldenrod
Solidago spp.
n
i
Hawkweed
Hieracium spp.
n
Horseweed
Conyza canadensis
n
Indiangrass
Sorghastrum nutans
n
Kentucky Bluegrass
Poa pratensis
n
i
Lance-leaved Coreopsis
Coreopsis lanceolata
n
48

Table 2.2. Continued
Common Name
Least Hop Clover
Lettuce
Lichen
Little Bluestem
Meadowgrass
Milkweed
Morning Glory
Moss
Musk Thistle
Mustard
Orchard Grass
Perennial Blue Flax
Perennial Ryegrass
Pigweed (Lamb‘s-quarters)
Plantain
Prickly Lettuce
Queen Anne's Lace
Ragweed
Red Clover
Redtop
Rush
Shasta Daisy ‗Alaska‘
Smooth Brome
Spiny-leaf Sow Thistle
Switchgrass
Thistle
Timothy
Wheat
White Clover
White Sweet Clover
Willow
Wood Sorrel
Yellow Prairie Coneflower
Yellow Sweet Clover
Number of Species Present

Plant Community Type
Scientific Name
Native
Introduced
Trifolium dubium
n
i
Lactuca spp.
n
i
Lichen spp.
n
Andropogon scoparius
n
Poa spp.
i
Asclepias spp.
n
i
Ipomoea spp.
i
Bryophyta spp.
n
i
Carduus nutans
n
Cruciferae spp.
i
Dactlyis glomerata
n
i
Linum perenne lewisii
n
Lolium perenne
n
Chenopodium album
n
Plantago spp.
n
i
Lactuca serriola
n
i
Daucus carota
n
i
Ambrosia spp.
i
Trifolium pratense
n
i
Agrostis gigantea
n
i
Juncus spp.
n
Chrysanthemum x superbum
n
Bromus inermis
n
i
Sonchus asper
i
Panicum virgatum
n
Cirsium spp.
n
i
Phleum pratense
n
i
Triticum aestivum
i
Trifolium repens
n
Melilotus alba
n
i
Salix spp.
n
Oxalis spp.
n
Ratibida columnifera
n
Melilotus officinalis
n
60
37

49

100
90
80
70
60
50
40
30
20
10
0

Outliers

Mean

Fig. 2.2. 2005 mid-summer (15 June – 14 July) vegetation variables of native and introduced vegetation of Conservation Reserve
Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan. For interpretation of the references to color in this and
all other figures, the reader is referred to the electronic version of this dissertation.
50

100
90
80
70
60
50
40

30
20
10
0

Outliers

Mean

Fig. 2.3. 2006 early summer (15 May - 14 June) vegetation variables of native and introduced vegetation of Conservation Reserve
Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan.

51

100
90
80
70
60
50
40
30
20
10
0

Outliers

Mean

Fig. 2.4. 2006 mid- summer (15 June - 14 July) vegetation variables of native and introduced vegetation of Conservation Reserve
Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan.
52

100
90
80
70
60
50
40

30
20
10
0

Outliers

Mean

Fig. 2.5. 2006 late summer (15 July - 14 August) vegetation variables of native and introduced vegetation of Conservation Reserve
Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan.

53

100

90
80
70
60
50
40
30
20
10
0

Outliers

Mean

Fig. 2.6. Native vegetation variables of CP23 and CP1 Conservation Reserve Enhancement Program (CREP) grasslands, Saginaw Bay
watershed, Michigan, mid-summer (15 June – 14 July) 2005.

54

100
90
80
70
60
50
40
30
20
10

0

Outliers

Mean

Fig. 2.7. Native vegetation variables of CP23 and CP1 Conservation Reserve Enhancement Program (CREP) grasslands, Saginaw Bay
watershed, Michigan, early summer (15 May – 14 June) 2006.

55

100
90
80
70
60
50
40
30
20
10

0

Outliers

Mean

Fig. 2.8. Native vegetation variables of CP23 and CP1 Conservation Reserve Enhancement Program (CREP) grasslands, Saginaw Bay
watershed, Michigan, mid-summer (15 June – 14 July) 2006.

56

100
90

80
70
60
50
40
30

20
10
0

Outliers

Mean

Fig. 2.9. Native vegetation variables of CP23 and CP1 Conservation Reserve Enhancement Program (CREP) grasslands, Saginaw Bay
watershed, Michigan, late summer (15 July – 14 August) 2006.
57

Table 2.3. Avian species observed in CP23 native fields, the introduced portion of CP1 fields, and CP1 whole fields, Conservation
Reserve Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan, 2005-2006. Avian densities (# individuals /ha
[SE]) are given for species with detection probability estimates ≥0.10. Species observed too rarely to support detection probability
modeling (i.e., observed in < 10% of surveys) or with detection probability estimates <0.10 are noted as present when observed in at least
one survey and as undetected when unobserved.
2005
Common Name

Scientific Name

Native/
CP23

Introduced

2006
CP1

Native/
CP23

Introduced

CP1

Grassland Specialist Species
Bobolink

A

Dolichonyx oryzivorus

0.38 [0.15] 1.33 [0.37]

2.28 [0.93]

0.60 [0.26] 0.68 [0.27]

0.35 [0.15]

Spiza americana

0.04 [0.03] 0.08 [0.07]

0.05 [0.06]

0.03 [0.01] 0.05 [0.03]

0.05 [0.03]

Sturnella magna

0.12 [0.09] 0.004 [0.004] 0.003 [0.003] 0.06 [0.04] 0.01 [0.01]

0.03 [0.02]

Ammodramus savannarum

0.05 [0.02] 0.00 [0.00]

0.00 [0.00]

0.08 [0.03] 0.01 [0.01]

0.01 [0.01]

Ammodramus henslowii

present

undetected

undetected

present

undetected

undetected

Eremophila alpestris

undetected

undetected

undetected

present

undetected

undetected

Circus cyaneus

undetected

present

undetected

present

undetected

undetected

Ring-necked Pheasant

Phasianus colchicus

0.15 [0.10] 0.03 [0.02]

0.07 [0.04]

0.12 [0.07] 0.13 [0.08]

0.07 [0.06]

Savannah Sparrow

Passerculus sandwichensis

0.98 [0.39] 0.37 [0.12]

0.55 [0.17]

2.14 [0.62] 0.94 [0.32]

1.22 [0.46]

Cistothorus platensis

0.01 [0.01] 0.00 [0.00]

present

0.07 [0.02] 0.06 [0.03]

0.03 [0.02]

Pooecetes gramineus

0.07 [0.03] 0.01 [0.01]

0.02 [0.01]

0.06 [0.01] 0.02 [0.01]

0.06 [0.04]

Dickcissel

AB

Eastern Meadowlark
Grasshopper Sparrow
Henslow‘s Sparrow

AB

AC

Horned Lark
Northern Harrier

Sedge Wren

A

Vesper Sparrow

A

58

Table 2.3 Continued.
2005
Common Name

Scientific Name

Native/
CP23

Introduced

2006
CP1

Native

Introduced

CP1

Other Species
American Goldfinch

Carduelis tristis

0.33 [0.13] 0.08 [0.03]

0.41 [0.32]

0.24 [0.08] 0.26 [0.07]

0.29 [0.08]

American Robin

Turdus migratorius

undetected

present

undetected

undetected

present

present

American Woodcock

Scolopax minor

undetected

undetected

undetected

present

undetected

undetected

Barn Swallow

Hirundo rustica

NA*

NA*

NA*

present

present

present

Brown Thrasher

Toxostoma rufum

undetected

undetected

undetected

undetected

present

present

Brown-headed Cowbird

Molothrus ater

present

undetected

undetected

present

present

present

Chipping Sparrow

Spizella passerina

undetected

undetected

undetected

undetected

present

present

Clay-colored Sparrow

Spizella pallida

present

present

present

0.05 [0.03] present

present

Common Yellowthroat

Geothlypis trichas

0.03 [0.02] 0.05 [0.04]

0.09 [0.06]

0.12 [0.04] 0.10 [0.06]

0.16 [0.12]

Eastern Kingbird

Tyrannus tyrannus

present

present

present

0.03 [0.01] 0.004 [0.003]

present

House Wren

Troglodytes aedon

present

present

undetected

undetected

undetected

undetected

Indigo Bunting

Passerina cyanea

present

present

present

undetected

present

present

59

Table 2.3 Continued.
2005
Native/
CP23

Scientific Name

Killdeer

Charadrius vociferus

present

undetected

undetected

present

undetected

undetected

Mallard

Anas platyrhynchos

present

undetected

undetected

present

undetected

undetected

Cistothorus palustris

undetected

undetected

undetected

present

undetected

undetected

Mourning Dove

Zenaida macroura

present

present

present

present

present

present

Northern Bobwhite Quail

Colinus virginianus

undetected

present

present

undetected

present

undetected

Northern Flicker

Colaptes auratus

present

undetected

undetected

present

undetected

present

Red-winged Blackbird

Agelaius phoeniceus

0.64 [0.26] 2.30 [0.29]

2.18 [0.62]

0.73 [0.19] 2.97 [0.36]

2.11 [0.26]

Song Sparrow

Melospiza melodia

1.54 [0.35] 0.46 [0.08]

1.47 [0.47]

1.89 [0.37] 0.76 [0.14]

1.12 [0.12]

Tree Swallow

Tachycineta bicolor

NA*

NA*

NA*

0.01 [0.01] present

0.06 [0.05]

White-crowned Sparrow

Zonotrichia leucophrys

undetected

undetected

undetected

undetected

undetected

present

Wild Turkey

Meleagris gallopavo

present

undetected

undetected

present

undetected

present

B

A

Michigan Priority Landbird (Knutson et al. 2001, Matteson et al. 2009)
Michigan Species of Special Concern
C
Michigan Endangered Species
* Barn Swallow and Tree Swallow were not included in the 2005 survey.
B

60

CP1

Native/
CP23

Common Name

Marsh Wren

Introduced

2006
Introduced

CP1

Table 2.4. Grassland bird species, with corresponding best-fitting detection models [with constant occupancy, ψ(·)] and detection
probability estimates, observed in 1) CP23 and CP1 Conservation Reserve Enhancement Program (CREP) grassland fields and 2)
CP23 native and CP1 introduced grassland plant communities, Saginaw Bay watershed, Michigan, 2005-2006. The lowest detection
probability estimate was reported when multiple detection probability estimates were generated.
CP23 and CP1 Whole CREP Fields

Native and Introduced Plant Communities

Common Name

Detection Model

p(·) [SE]

Detection Model

p(·) [SE]

Bobolink

p(Ordinal Date)

0.63 [0.06]

p(.)

0.75 [0.08]

Dickcissel

p(.)

0.35 [0.06]

p(Ordinal Date)

0.12 [0.07]

Eastern Meadowlark

p(CREP Field Type)

0.13 [0.07]

p(.)

0.34 [0.18]

Grasshopper Sparrow

p(.)

0.31 [0.05]

p(.)

0.47 [0.16]

Ring-necked Pheasant

p(CREP Field Type, Ordinal Date)

0.17 [0.05]

p(.)

0.28 [0.09]

Savannah Sparrow

p(Ordinal Date)

0.81 [0.05]

p(.)

0.84 [0.04]

Sedge Wren

p(Ordinal Date)*

0.12 [0.07]

p(Plant Community Type)

0.14 [0.16]

Vesper Sparrow

p(.)

0.21 [0.04]

p(.)

0.18 [0.09]

*Sedge Wren data from 2006 only

61

Detection Probability (%)

1.00
0.90
0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
1-14
June

15-30
June

1-14
July

15-31 1-14
July August

15-31
May

2005

1-14
June

15-30
June

1-14
July

15-31 1-14
July August

2006

Fig. 2.10. Relationship between Bobolink detection probability and ordinal date in CP23 and CP1 Conservation Reserve
Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan, 2005-2006. Error bars indicate 95% confidence
intervals.

62

1.00
0.90
Detection Probability (%)

0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
CP23

CP1
CREP Field Type

Fig. 2.11. Relationship between Eastern Meadowlark detection probability and
Conservation Reserve Enhancement Program (CREP) administrative field type in CP23
and CP1 CREP grasslands, Saginaw Bay watershed, Michigan, 2005 and 2006. Error
bars indicate 95% confidence intervals.

63

CP23 Probability of Detection
CP1 Probability of Detection

1.00

Detection Probability (%)

0.90
0.80
0.70
0.60

0.50
0.40
0.30
0.20
0.10
0.00
1-14
June

15-30
June

1-14
July

15-31 1-14
July August

15-31
May

1-14
June

15-30
June

1-14
July

15-31 1-14
July August

2006

2005

Fig. 2.12. Relationship between Ring-necked Pheasant detection probability, field type, and ordinal date in CP23 and CP1
Conservation Reserve Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan, 2005 and 2006. Error bars
indicate 95% confidence intervals.

64

1.00
Detection Probability (%)

0.90
0.80
0.70
0.60

0.50
0.40
0.30
0.20
0.10
0.00
1-14
June

15-30
June

1-14
July

15-31 1-14
July August

15-31
May

2005

1-14
June

15-30
June

1-14
July

15-31 1-14
July August

2006

Fig. 2.13. Relationship between Savannah Sparrow detection probability and ordinal date in CP23 and CP1 Conservation Reserve
Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan, 2005-2006. Error bars indicate 95% confidence
intervals.

65

1.00
0.90

Detection Probability (%)

0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10

0.00
15-31
May

1-14
June

15-30
June

1-14
July

15-31
July

1-14
August

2006

Fig. 2.14. Relationship between Sedge Wren detection probability and ordinal date in
CP23 and CP1 Conservation Reserve Enhancement Program (CREP) grasslands,
Saginaw Bay watershed, Michigan, 2006. Error bars indicate 95% confidence intervals.

66

1.00
0.90

Detection Probability (%)

0.80
0.70
0.60

0.50
0.40
0.30
0.20

0.10
0.00
1-14
June

15-30
June

1-14
July

15-31
July

15-31
May

2005

1-14
June

15-30
June

1-14
July

15-31
July

2006

Fig. 2.15. Relationship between Dickcissel detection probability and ordinal date in CP23 native and CP1 introduced plant
communities of Conservation Reserve Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan, 2005-2006.
Error bars indicate 95% confidence intervals.
67

1.00
0.90

Detection Probability (%)

0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
Native

Introduced

Grassland Plant Community Type
Fig. 2.16. Relationship between Sedge Wren detection probability and grassland type in
CP23 native and CP1 introduced plant communities of Conservation Reserve
Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan, 2005 and
2006. Error bars indicate 95% confidence intervals.

68

6

Density (# individuals / ha)

5

4

3

2

1

0
Native / CP23
2005

Introduced 2005

CP1 2005

Native / CP23
2006

Introduced 2006

CP1 2006

Plant Community or Whole Field Type
Outliers

Mean

Fig. 2.17. Savannah Sparrow density (# individuals / ha) in native/CP23 whole fields, introduced vegetation of CP1 fields, and CP1
whole fields of Conservation Reserve Enhancement Program (CREP) grasslands in Michigan‘s Saginaw Bay watershed, 2005 and
2006.

69

Density (# individuals / ha)

5.0

4.0

3.0

2.0

1.0

0.0
Native / CP23
2005

Introduced 2005

CP1 2005

Native / CP23
2006

Introduced 2006

CP1 2006

Plant Community or Whole Field Type
Outlier

Mean

Fig. 2.18. Bobolink density (# individuals / ha) in native/CP23 whole fields, introduced vegetation of CP1 fields, and CP1 whole
fields of Conservation Reserve Enhancement Program (CREP) grasslands in Michigan‘s Saginaw Bay watershed, 2005 and 2006.

70

0.5

Density (# individuals / ha)

0.4

0.3

0.2

0.1

0.0
Native / CP23
2005

Introduced 2005

CP1 2005

Native / CP23
2006

Introduced 2006

CP1 2006

Plant Community or Whole Field Type
Outliers

Mean

Fig. 2.19. Dickcissel density (# individuals / ha) in native/CP23 whole fields, introduced vegetation of CP1 fields, and CP1 whole
fields of Conservation Reserve Enhancement Program (CREP) grasslands in Michigan‘s Saginaw Bay watershed, 2005 and 2006.

71

Density (# individuals / ha)

0.6
0.5
0.4
0.3
0.2
0.1

0.0
Native / CP23
2005

Introduced 2005

CP1 2005

Native / CP23
2006

Introduced 2006

CP1 2006

Plant Community or Whole Field Type
Outliers

Mean

Fig. 2.20. Eastern Meadowlark density (# individuals / ha) in native/CP23 whole fields, introduced vegetation of CP1 fields, and CP1
whole fields of Conservation Reserve Enhancement Program (CREP) grasslands in Michigan‘s Saginaw Bay watershed, 2005 and
2006.
72

0.25

Density (# individuals / ha)

0.20

0.15

0.10

0.05

0.00
Native / CP23
2005

Introduced 2005

CP1 2005

Native / CP23
2006

Introduced 2006

CP1 2006

Plant Community or Whole Field Type
Outliers

Mean

Fig. 2.21. Grasshopper Sparrow density (# individuals / ha) in native/CP23 whole fields, introduced vegetation of CP1 fields, and
CP1 whole fields of Conservation Reserve Enhancement Program (CREP) grasslands in Michigan‘s Saginaw Bay watershed, 2005
and 2006.
73

0.7

Density (# individuals / ha)

0.6
0.5
0.4
0.3
0.2

0.1
0.0
Native / CP23
2005

Introduced 2005

CP1 2005

Native / CP23
2006

Introduced 2006

CP1 2006

Plant Community or Whole Field Type
Outliers

Mean

Fig. 2.22. Ring-necked Pheasant density (# individuals / ha) in native/CP23 whole fields, introduced vegetation of CP1 fields, and
CP1 whole fields of Conservation Reserve Enhancement Program (CREP) grasslands in Michigan‘s Saginaw Bay watershed, 2005
and 2006.
74

Density (# individuals / ha)

0.3

0.2

0.1

,

0.0
Native / CP23
2005

Introduced 2005

Native / CP23
2006

Introduced 2006

CP1 2006

Plant Community or Whole Field Type
Outlier

Mean

Fig. 2.23. Sedge Wren density (# individuals / ha) in native/CP23 whole fields, introduced vegetation of CP1 fields, and CP1 whole
fields of Conservation Reserve Enhancement Program (CREP) grasslands in Michigan‘s Saginaw Bay watershed, 2005 and 2006.

75

0.25

Density (# individuals / ha)

0.20

0.15

0.10

0.05

0.00
Native / CP23
2005

Introduced 2005

CP1 2005

Native / CP23
2006

Introduced 2006

CP1 2006

Plant Community or Whole Field Type

Outlers

Mean

Fig. 2.24. Vesper Sparrow density (# individuals / ha) in native/CP23 whole fields, introduced vegetation of CP1 fields, and CP1
whole fields of Conservation Reserve Enhancement Program (CREP) grasslands in Michigan‘s Saginaw Bay watershed, 2005 and
2006.
76

Table 2.5. Grassland bird species, with estimated Michigan breeding territory size range (ha) and estimated number of breeding
territories in CP23 and CP1 Conservation Reserve Enhancement Program (CREP) fields and CP1 introduced grassland plant
communities, Saginaw Bay watershed, Michigan, 2005-2006.
Number of Territories (Range) In:
CP23 Native
A
Grasslands

CP1 Whole
B
Fields

CP1 Introduced
C
Grasslands

Martin and Gavin 1995

4 - 40

4 - 49

3 - 35

0.3 - 1.1

Temple 2002

6 - 66

7 - 80

5 - 57

Eastern Meadowlark

2.8 - 3.2

Lanyon 1995

2-7

2-9

2-6

Grasshopper Sparrow

0.85 - 1.4

Vickery 1996

5 - 23

5 - 28

4 - 20

Ring-necked Pheasant 4 – 8

Sargent and Carter 1999

1-5

1-6

1-4

Savannah Sparrow

0.11 - 0.86

Wheelwright and Rising 2008

8 - 180

9 - 218

6 - 155

Sedge Wren

0.13 - 0.36

Herkert et al. 2001

19 - 152

20 - 184

15 - 131

Vesper Sparrow

2.59 - 8.19

Jones and Cornely 2002

1-8

1-9

1-7

Grassland Bird
Species

Estimated Breeding
Territory Size Range (ha)

Citation

Bobolink

0.49 - 2.0

Dickcissel

A
B
C

CP23 native grasslands range from 6.9 -19.8 ha in size (mean [SE] = 12.6 [1.9] ha, median = 11.8 ha).
CP1 whole fields range from 7.3 – 23.9 ha in size (mean [SE] = 12.3 [2.1] ha, median =10.9).
CP1 introduced grasslands range from 5.5 - 17.0 ha in size (mean [SE] = 9.4 [1.5] ha, median = 8.0).

77

APPENDIX 2.2: DETECTION PROBABILITY MODELING AND
DENSITY COMPARISON RESULTS FOR
AVIAN SPECIES OTHER THAN GRASSLAND SPECIALISTS

78

RESULTS
A total of 34 avian species were observed in CREP study fields during the 2 years
of this study (Table 2.3). Based upon the grassland bird list developed for the BBS by
the U.S. Geographic Survey (Sauer et al. 1995), 23 of the observed species were
identified as species other than grassland specialists (Table 2.3).
Avian detection probability modeling
Detection probability estimates were generated for non-grassland avian species in
native and introduced CREP grassland plant communities. Fifteen non-grassland species
were observed too rarely (i.e., observed in less than 10% of surveys) to estimate their
detection probability: American Robin (Turdus migratorius), American Woodcock
(Scolopax minor), Brown Thrasher (Toxostoma rufum), Brown-headed Cowbird
(Molothrus ater), Chipping Sparrow (Spizella passerina), House Wren (Troglodytes
aedon), Indigo Bunting (Passerina cyanea), Killdeer (Charadrius vociferus), Mallard
(Anas platyrhynchos), Marsh Wren (Cistothorus palustris), Mourning Dove (Zenaida
macroura), Northern Bobwhite Quail (Colinus virginianus), Northern Flicker (Colaptes
auratus), White-crowned Sparrow (Zonotrichia leucophrys), and Wild Turkey (Meleagris
gallopavo). Barn Swallow (Hirundo rustica), was observed frequently enough to perform
detection probability modeling, but no successful detection models were generated.
Successful detection models were obtained for Eastern Kingbird (Tyrannus tyrannus) only
for CP23 fields and the introduced vegetation of CP1 fields.
To enable reliable comparisons of non-grassland bird density in 1) CP23 and CP1
fields (i.e., the program administrative-level comparison) and in 2) CP23 fields and CP1
introduced plant communities (i.e., the ecological-level comparison), the detection

79

probability, p, of each avian species was estimated at both scales. Successful detection
probability modeling occurred for 7 avian species: American Goldfinch (Carduelis
tristis), Clay-colored Sparrow (Spizella pallida), Common Yellowthroat (Geothlypis
trichas), Eastern Kingbird, Red-winged Blackbird (Agelaius phoeniceus), Song Sparrow
(Melospiza melodia), and Tree Swallow (Tachycineta bicolor) (Table 2.6). Adequate
detection probability estimates (i.e., p ≥ 0.10) were achieved for all species, and ranged
from 0.19 to 0.90 in CP23 and CP1 fields and from 0.17 to 0.84 in CP23 fields and
introduced vegetation (Table 2.6). Adequate detection probability estimates were
obtained only when a subset of the data was used for Clay-colored Sparrow (native 2006
data) and Eastern Kingbird (2006 data) (Table 2.6). The linear effects of the cover type
(native or introduced vegetation) and/or standardized ordinal date were included in the
best detection models for American Goldfinch, Clay-colored Sparrow, Common
Yellowthroat, Red-winged Blackbird, and Song Sparrow (Table 2.6).
Avian density comparisons
Avian species were categorized as reliably detected (i.e., p ≥0.10) or unreliably
detected (i.e., p could not be estimated or p <0.10) in each CREP field/grassland type.
Avian densities were reported (Table 2.3) and compared (Figures 2.25-2.30) only for
those species that were reliably detected. Species that were unreliably detected were
designated as present if observed in a field/grassland type and as undetected if not
observed (i.e., species that were unreliably detected actually may have been present when
unobserved; Table 2.3).
Sixteen non-grassland avian species were observed in 2005; 14 species were
recorded in CP23 fields, 11 species were found in introduced vegetation, and 9 species

80

were detected in CP1 fields (Table 2.3). None of the 4 reliably detected species were
unique to any field/vegetation type. In 2006, a total of 21 non-grassland species were
observed, 16 species in CP23 fields, 15 species in introduced habitat, and 17 species in
CP1 fields (Table 2.3). None of the 5 reliably detected species were unique to any
field/vegetation type. Three species were detected in the native portions of CP1 fields
that were unobserved in the introduced portions of CP1 fields: Northern Flicker, Whitecrowned Sparrow, and Wild Turkey in 2006 (Table 2.3). Two species, American
Goldfinch (2005) and Northern Bobwhite Quail (2006) were detected in the introduced
vegetation of CP1 fields but were recorded as undetected for CP1 whole fields because
data were used from only 5 of the 7 CP1 whole fields (i.e., data from fields where the
native plant community portion did not establish successfully were omitted (Table 2.3)).
Among adequately detected species (i.e., p ≥ 0.10), the most abundant non-grassland
birds during both years of the study were Song Sparrow and Red-winged Blackbird, with
the highest mean (Table 2.3) and median densities (Figures 2.25-2.30) in both native and
introduced vegetation and in CP23 and CP1 whole fields (see range, median, and mean
information below).
Avian density comparisons were conducted for species that were adequately
detected (i.e., p ≥ 0.10): American Goldfinch, Clay-colored Sparrow, Common
Yellowthroat, Eastern Kingbird, Red-winged Blackbird, Song Sparrow, and Tree
Swallow. Boxplots were used to compare avian densities in 1) CP23 and CP1 fields and
in 2) CP23 fields and the introduced vegetation of CP1 fields (Figures 2.25-2.30).
American Goldfinch density was larger in CP23 fields than in introduced
vegetation in 2005, and no differences in density were observed in 2006 (Figure 2.25). In

81

2005, American Goldfinch density ranged from 0.05 – 0.89 individuals/ha in CP23 fields
(median = 0.20, mean = 0.33, SE = 0.13) and from 0 – 0.20 in CP1 fields (median = 0.08,
mean = 0.08, SE = 0.03; Figure 2.25, Table 2.3). Clay-colored Sparrow was reliably
detected only in 2006 in CP23 fields, so density comparisons were not possible. Clay
Colored density ranged from 0 – 0.22 individuals/ha (median = 0.02, mean = 0.05, SE =
0.03) in CP23 fields in 2006. Common Yellowthroat density in CP1 fields was more
variable but not clearly larger than in CP23 fields in 2005; in 2006, Common
Yellowthroat density was larger in CP23 fields than in the introduced vegetation of CP1
fields or in CP1 whole fields (Figure 2.26). In 2006, Common Yellowthroat density
ranged from 0.01 – 0.28 individuals/ha in CP23 fields (median = 0.08, mean = 0.12, SE =
0.04), from 0 – 0.40 individuals/ha in introduced vegetation (median = 0.02, mean = 0.10,
SE = 0.06), and from 0 – 0.62 individuals/ha in CP1 fields (median = 0.03, mean = 0.16,
SE = 0.12; Figure 2.26, Table 2.3). In 2006, Eastern Kingbird density was larger in CP23
fields than in introduced vegetation (Figure 2.27). In 2006, Eastern Kingbird density
ranged from 0 – 0.05 individuals/ha in CP23 fields (median = 0.02, mean = 0.03, SE =
0.01) and from 0 – 0.02 individuals/ha in introduced vegetation (median = 0, mean =
0.004, SE = 0.003; Figure 2.27, Table 2.3). Red-winged Blackbird density was smaller in
CP23 fields than in CP1 introduced vegetation or in CP1 whole fields during both years
of the study (Figure 2.28). In 2005, Red-winged Blackbird density ranged from 0.02 –
1.48 individuals/ha in CP23 fields (median = 0.22, mean = 0.64, SE = 0.26), from 0.95 –
3.37 individuals/ha in introduced vegetation (median = 2.6, mean = 2.30, SE = 0.29), and
from 0.93 – 4.06 individuals/ha in CP1 fields (median = 1.6, mean = 2.18, SE = 0.62;
Figure 2.28, Table 2.3). In 2006, Red-winged Blackbird density ranged from 0.02 – 1.43

82

individuals/ha in CP23 fields (median = 0.67, mean = 0.73, SE = 0.19), from 1.67 – 4.55
individuals/ha in introduced vegetation (median = 2.63, mean = 2.97, SE = 0.36), and
from 1.17 – 2.70 individuals/ha in CP1 fields (median = 2.19, mean = 2.11, SE = 0.26;
Figure 2.28, Table 2.3). In contrast, Song Sparrow density was larger in CP23 fields than
in introduced vegetation during both years of the study (Figure 2.29). In 2005, Song
Sparrow density ranged from 0.80 – 3.01 individuals/ha in CP23 fields (median = 1.12,
mean = 1.54, SE = 0.35) and from 0.18 – 0.75 individuals/ha in introduced vegetation
(median = 0.37, mean = 0.46, SE = 0.08; Figure 2.29, Table 2.3). In 2006, Song Sparrow
density ranged from 0.80 – 3.36 individuals/ha in CP23 fields (median = 1.38, mean =
1.89, SE = 0.37) and from 0.14 – 1.22 individuals/ha in introduced vegetation (median =
0.72, mean = 0.76, SE = 0.14; Figure 2.29, Table 2.3). No differences in Tree Swallow
density in CP23 or CP1 fields occurred in 2006, the only year reliable density
information for the species was available (Figure 2.30).

83

Table 2.6. Non-grassland avian species, with corresponding best-fitting detection models [with constant occupancy, ψ(·)] and
detection probability estimates, observed in 1) CP23 and CP1 Conservation Reserve Enhancement Program (CREP) grassland fields
and 2) CP23 native and CP1 introduced grassland plant communities, Saginaw Bay watershed, Michigan, 2005-2006. The lowest
detection probability estimate was reported when multiple detection probability estimates were generated.
CP23 and CP1 Whole CREP Fields

Native and Introduced Plant Communities

Common Name

Detection Model

p(·) [SE]

American Goldfinch

p(CREP Field Type)

0.56 [0.06] p(Grassland Type)

0.42 [0.06]

Clay-colored Sparrow,
native 2006 data

p(Ordinal Date)

0.21 [0.14] p(Ordinal Date)

0.21 [0.14]

Common Yellowthroat

p(Ordinal Date)

0.23 [0.06] p(Ordinal Date)

0.20 [0.05]

p(.)

0.17 [0.07]

Eastern Kingbird, 2006 data

Detection Model

p(·) [SE]

Red-winged Blackbird

p(CREP Field Type, Ordinal Date)

0.56 [0.09] p(Grassland Type, Ordinal Date)

0.56 [0.09]

Song Sparrow

p(CREP Field Type)

0.90 [0.03] p(Grassland Type)

0.84 [0.04]

Tree Swallow

p(.)

0.19 [0.09]

84

1.50

Density (# individuals / ha)

1.25
1.00
0.75
0.50
0.25
0.00
Native / CP23
2005

Introduced 2005

CP1 2005

Native / CP23
2006

Introduced 2006

CP1 2006

Plant Community or Whole Field Type
Outliers

Mean

Fig. 2.25. American Goldfinch density (# individuals / ha) in native/CP23 whole fields, introduced vegetation of CP1 fields, and CP1
whole fields of Conservation Reserve Enhancement Program (CREP) grasslands in Michigan‘s Saginaw Bay watershed, 2005 and
2006.

85

0.60
0.55
Density (# individuals / ha)

0.50

0.45
0.40
0.35

0.30
0.25
0.20
0.15
0.10
0.05
0.00
Native / CP23
2005

Introduced 2005

CP1 2005

Native / CP23
2006

Introduced 2006

CP1 2006

Plant Community or Whole Field Type
Outliers

Mean

Fig. 2.26. Common Yellowthroat density (# individuals / ha) in native/CP23 whole fields, introduced vegetation of CP1 fields, and
CP1 whole fields of Conservation Reserve Enhancement Program (CREP) grasslands in Michigan‘s Saginaw Bay watershed, 2005
and 2006.

86

Density (# individuals / ha)

0.075

0.050

0.025

0.000
Native / CP23
2006

Introduced 2006

Plant Community Type

Outlier

Mean

Fig. 2.27. Eastern Kingbird density (# individuals / ha) in native/CP23 whole fields and introduced vegetation of CP1 fields of
Conservation Reserve Enhancement Program (CREP) grasslands in Michigan‘s Saginaw Bay watershed, 2006.

87

Density (# individuals / ha)

5.0

4.0

3.0

2.0

1.0

0.0
Native / CP23
2005

Introduced 2005

CP1 2005

Native / CP23
2006

Introduced 2006

CP1 2006

Plant Community or Whole Field Type
Outliers

Mean

Fig. 2.28. Red-winged blackbird density (# individuals / ha) in native/CP23 whole fields, introduced vegetation of CP1 fields, and
CP1 whole fields of Conservation Reserve Enhancement Program (CREP) grasslands in Michigan‘s Saginaw Bay watershed, 2005
and 2006.

88

4.5

Density (# individuals / ha)

4.0
3.5
3.0
2.5

2.0
1.5
1.0

0.5
0.0
Native / CP23
2005

Introduced 2005

CP1 2005

Native / CP23
2006

Introduced 2006

CP1 2006

Plant Community or Whole Field Type
Outlier

Mean

Fig. 2.29. Song Sparrow density (# individuals / ha) in native/CP23 whole fields, introduced vegetation of CP1 fields, and CP1 whole
fields of Conservation Reserve Enhancement Program (CREP) grasslands in Michigan‘s Saginaw Bay watershed, 2005 and 2006.

89

Density (# individuals / ha)

0.25

0.20

0.15

0.10

0.05

0.00
Native / CP23
2006

CP1 2006

Whole Field Type
Outlier

Mean

Fig. 2.30. Tree Swallow density (# individuals / ha) in native/CP23 and whole fields of Conservation Reserve Enhancement Program
(CREP) grasslands in Michigan‘s Saginaw Bay watershed, 2006.
90

APPENDIX 2.3: COMPARISON OF AVIAN PRODUCTIVITY WITHIN CP23
NATIVE AND CP1 INTRODUCED PLANT COMMUNITIES OF MICHIGAN
CONSERVATION RESERVE ENHANCEMENT PROGRAM (CREP)
GRASSLANDS

91

METHODS
To investigate the source/sink status of CREP study fields, avian productivity and
nesting success in CREP lands were examined. All study fields were searched for nests
one time in June and once in July, 2006. Due to the large size of several study fields (i.e.,
>20 ha), the nest search area in all 14 study fields was equal to the smallest study field
size, or approximately 6.8 hectares. Nest searches were conducted by 3 – 6 observers
walking 3 – 5 m abreast while traversing the nest search areas. As the observers traveled,
they swept the top of the vegetation with yardsticks to encourage sitting birds to flush
from their nests. Nests were revisited every 2 – 3 days until the young fledged or nests
were abandoned or destroyed. To minimize the chance that predators would be attracted
to nests due to researcher activity, nests were marked with flags placed 5 m north of
actual nest location, and different paths were taken to nests during each visit. Nest
location was mapped using a handheld Global Positioning System (GPS) unit which
recorded the latitude and longitude coordinates of nest sites and facilitated navigation to
nests during revisits. Nests discovered while performing bird surveys or vegetation
sampling were also revisited and data were recorded as for any other nest. Avian species
and nest outcome were recorded for each nest. Successful nests were defined as those
that produced at least one fledgling. If no fledglings were noted in the area of an empty
nest, fledglings were assumed to be produced and the nest outcome was recorded as
successful if 1) nestlings were close to fledgling status at the time of the last nest visit and
2) there was no evidence of predation (i.e., highly disturbed or destroyed nest structure or
the presence of nestling remains). Unsuccessful nests were those from which eggs or
young disappeared and fledglings could not be assumed to have been produced. Active

92

nests were classified as those that produced at least one egg or hatchling within the time
period of the study. Percent nesting success for an avian species was calculated as the
number of successful nests divided by the number of active nests after the number of
nests with unknown outcome was subtracted.

% Nesting Success =

# Successful Nests
.
(# Active Nests – # Nests with Unknown Outcome)

RESULTS
Intensive grassland bird nest search efforts in CREP fields were unsuccessful,
largely due to the extremely cryptic nature of the nests of most grassland bird species.
Lack of visual cues for nests was often compounded by the tendency of adult birds to flee
on foot beneath the plant canopy prior to or instead of flying when flushed from the nest,
making it difficult to pinpoint nest locations. Unknown nest outcome was the result of
unsuccessful efforts to relocate highly cryptic nests, despite the use of flagging and a
GPS with nest location coordinates. Nest relocation efforts, as well as efforts to initially
locate a nest that was assumed to be in a particular area due to adult bird behaviors such
as food transport, were abandoned if nest searches could not be performed without
endangering the nest by severely disturbing the nearby vegetation, which would make
direct damage to the nest and/or attraction of predators to the nesting area likely.
During 2006, a total of 28 individual active nests of seven species were found and
monitored in the 14 nest search areas (Table 2.7). The vast majority of nests observed
(i.e., 21) were those of Red-winged Blackbirds. Two Song Sparrow nests were also
discovered, along with 1 nest of each of the following species: Ring-necked Pheasant,
Savannah Sparrow, Vesper Sparrow, American Goldfinch, and Mallard. Ring-necked

93

Pheasant, Vesper Sparrow, and Song Sparrow nests all occurred in native vegetation,
while Savannah Sparrow, American Goldfinch, Mallard, and Red-wing Blackbird nests
were observed in introduced vegetation (Table 2.7).
The nesting success of American Goldfinch could not be defined because the
single active nest observed was not relocated and nest outcome was unknown. The
percent nesting success of all other species could be calculated, but so few nests were
discovered for most species that this statistic was only informative for Red-winged
Blackbird. Sixty-nine percent of active Red-wing Blackbird nests with known outcome
produced at least 1 fledgling (Table 2.7). The Ring-necked Pheasant and Mallard nests
that were observed were unsuccessful; both nests were destroyed, apparently by predators
(Table 2.7). The single active Savannah Sparrow nest was successful in producing at
least one fledgling, as was the single active Vesper Sparrow nest observed and the 2
active Song Sparrow nests that were monitored (Table 2.7).
DISCUSSION
Given unsuccessful efforts to quantify avian productivity, it is impossible to
evaluate the sink/source status of the CREP grasslands in this study. However, there is
increasing evidence that farmland set-aside lands are benefiting grassland bird species at
the population level (e.g., Veech 2006, Herkert 2009). In a study of the response of avian
populations in the north-central region of the U.S. to CRP farmland set-aside programs,
Herkert (2009) found that avian species with abundances that were much higher in CRP
fields than in row crop fields tended to have declining populations prior to CRP
establishment (1966-86) and tended to have an increasing population trend in the years
after CRP establishment (1987-2007), a pattern that appeared to be strongest for the

94

group of grassland obligate species in the study. Herkert (2009) also found that other
avian species, including some grassland species, showed population trends that declined
more steeply in the years after CRP establishment. However it is impossible to predict
what the trend in population size of any species would have been without the effect of
farmland set-aside programs; perhaps population trends would have been even more
negative in the absence of set-aside fields. Veech (2006) compared the landscapes
occupied by increasing and decreasing grassland-nesting avian species in 18 states in the
Midwest and Great Plains regions and found that CRP grasslands were beneficial to
grassland bird population growth. Even small (3-142 ha) grassland fragments have been
found to have a positive effect on the population fecundity of some grassland birds, such
as Dickcissel (estimated breeding territory size = 0.3 – 1.1 ha; Temple 2002) and Eastern
Meadowlark (estimated breeding territory size = 2.8 – 3.2 ha; Lanyon 1995) (Walk et al.
2010), indicating that the loss of small habitat patches such as the CREP grasslands in
this study could have negative, perhaps cumulative, impacts on the populations of
grassland birds, and emphasizing the importance of maintaining farmland set-aside
grasslands such as those in this study.

95

Table 2.7. Active avian nests observed, nest outcome, and vegetation type of the nest location in Conservation Reserve Enhancement
Program (CREP) whole field grasslands, Saginaw Bay watershed, Michigan, 2006.
Avian Species

Total active nests

Successful nests

Unsuccessful nests

Unknown nest outcome

Grassland Species
A

Ring-necked Pheasant

1

Savannah Sparrow

1

Vesper Sparrow

1

B

A

A

1

0
B

1

0

0

0

1

0

0

0

0

1

0

1

A

Non-grassland Species
B

American Goldfinch

1

Mallard

1

B
B

21

Red-winged Blackbird

A

2

Song Sparrow
Total
A
B

28

B

B

B

B

0
B

5 *

5

2

0

0

15

7*

6

11

A

Nest(s) occurred in native vegetation.
Nest(s) occurred in introduced vegetation.

*A total of two nests were destroyed when landowner/manager mowed within and between thistle patches.

96

APPENDIX 2.4: COMPARISON OF AVIAN SPECIES PRESENCE IN
MICHIGAN CONSERVATION RESERVE ENHANCEMENT PROGRAM
(CREP) AND CONSERVATION RESERVE PROGRAM (CRP)
INTRODUCED GRASSLAND VEGETATION

97

METHODS
Results from this 2005-06 study of avian populations of Conservation Reserve
Enhancement Program (CREP) grasslands were compared to those from a study of avian
populations of Conservation Reserve Program (CRP) grasslands performed in 1991-92
(Millenbah 1993). CRP introduced grassland plantings in Gratiot County, Michigan, were
studied in 1991 (n = 12, 7.3-19.8 ha) and 1992 (n = 19, including 10 fields also studied in
1991, 7.7-19.8 ha; Millenbah 1993). Vegetation sampling was performed at 20 m
intervals along six 100 m long transects in each CRP field in April just after snow melt, in
May, and in July (Millenbah 1993). Structural variables of the CRP vegetation were
visually estimated using a 50 x 50 cm Daubenmire frame (Daubenmire 1959). Avian
surveys in CRP fields were conducted using line transect sampling and the entire field area
was observed as previously described for CREP fields (Chapter 2) (Millenbah 1993).
Detection probability modeling was not performed during the CRP study, so the
accuracy of absence records of species observed at that time is uncertain, as is the
accuracy of absence records for species without adequate detection probability estimates,
p, (i.e., p ≥ 0.10) from the CREP study. However, no such uncertainly is associated with
the presence records of species from either study, so patterns of avian presence in the two
studies were compared.
RESULTS
Means of mid-summer vegetation variables from same-aged CREP and CRP
introduced grasslands were compared (Table 2.8). In CREP vegetation, the plant canopy
composed the highest percentage of field area, followed by a smaller proportion of litter
cover and even less bare ground, a pattern that also appeared to occur in CRP fields (Table

98

2.8). Within the canopy, live vegetation was more abundant than dead vegetation in both
types of grasslands (Table 2.8). Grasses tended to be more abundant than forbs in the live
canopy of CRP fields, while grasses and forbs in CREP vegetation grasses tended to be
equally abundant (Table 2.8). Although similar in both grassland types, litter depth in
CREP vegetation tended to be greater than in CRP fields, while horizontal cover tended to
be greater in CRP fields, possibly due to the inclusion of sweet clover in CRP planting
mixes (Table 2.8).
Presence records for all avian species observed in CRP and CREP introduced
grasslands were summarized, along with avian density information for species that were
reliably detected in the CREP study (Table 2.9). A total of 10 grassland specialist species
were observed in the introduced vegetation of both the CRP and CREP studies: Bobolink,
Dickcissel, Eastern Meadowlark, Grasshopper Sparrow, Horned Lark, Northern Harrier,
Ring-necked Pheasant, Savannah Sparrow, Sedge Wren, and Vesper Sparrow. All 10
species were observed in introduced CRP introduced fields, while 9 species, all except
Horned Lark, were observed in introduced CREP introduced vegetation (Table 2.9). A
total of 31 non-grassland species were observed in the introduced vegetation of both the
CRP and CREP studies; 14 species were observed in both studies, 3 species were observed
only in the introduced CREP vegetation, and 14 species were observed only in the CRP
fields studied in 1991-92 (Table 2.9).

99

Table 2.8. Mean [SE] vegetative characteristics of 1.) introduced Conservation Reserve Enhancement Program (CREP) grasslands,
Saginaw Bay watershed, Michigan, mid-summer (15 June – 14 July) 2005 and mid-summer 2006 (n = 7), and 2.) Conservation
Reserve Program (CRP) grasslands, Gratiot County, Michigan, July 1991 and July 1992 (n = 3). CREP fields were planted in 2002
and CRP fields were planted in 1988.
CREP Introduced Grasslands
A
(5.5 – 17.1 ha)
Characteristic

CRP Introduced Grasslands
BC
(8.1 – 19.8 ha)

2005

2006

% Bare Ground

2.5 [1.0]

1.1 [0.6]

~0.5

% Litter Cover

9.6 [2.9]

35.2 [5.1]

~0.5

% Total Canopy

87.9 [3.8]

63.8 [5.4]

99.0 [1.8]

92.5 [4.5]

% Dead Canopy

2.3 [1.6]

0.3 [0.1]

6.6 [0.6]

10.5 [0.3]

% Live Canopy

85.5 [3.7]

63.4 [5.5]

92.4 [2.9]

81.9 [4.8]

42.7 [5.0]

27.0 [1.5]

0.0 [0.0]

0.0 [0.0]

42.8 [6.4]

36.4 [6.1]

~36.5

Litter Depth (cm)

5.8 [0.6]

5.6 [0.6]

NA

3.4 [0.8]

Horizontal Cover (dm)

6.2 [0.5]

8.2 [0.4]

10.8 [0.1]

9.0 [0.4]

% Live Grass
% Live Woody
% Live Forb

A
B
C

1991

~55.9
0.0 [0.0]

1992
~3.2
4.3 [3.8]

~51.1
0.0 [0.0]
~30.8

Planting mixture all fields: 23% Timothy, 23% Orchardgrass, 27% Alfalfa, 27% Red Clover; planted 12.4 kg/ha.
Planting mixture 2 fields: 22% Timothy, 34% Orchardgrass, 22% Alfalfa, 22% White Sweet Clover; planted 10 kg/ha.
Planting mixture 1 field: 22% Timothy, 45% Orchardgrass, 22% Alfalfa, 11% White Sweet Clover; planted 10 kg/ha.

100

Table 2.9. Avian species observed in 1.) introduced vegetation of Conservation Reserve Enhancement Program (CREP) grasslands,
Saginaw Bay watershed, Michigan, 2005-2006, and 2.) introduced vegetation of Conservation Reserve Program (CRP) grasslands,
Gratiot County, Michigan, 1991-1992. Avian densities (# individuals/ha) [SE] are given for species with successful detection
probability modeling (p≥0.10). In CREP vegetation, species observed too rarely to support detection probability modeling (i.e.,
observed in < 10% of surveys) or with detection probability estimates <0.10 are noted as present when observed in at least one survey
and as undetected when unobserved. Detection probability modeling was not performed for avian species observed in CRP vegetation.
Common Name

Scientific Name

2005

2006

1991

1992

Dolichonyx oryzivorus

1.33 [0.37]

0.68 [0.27]

present

present

Spiza americana

0.08 [0.07]

0.05 [0.03]

undetected

present

Sturnella magna

0.004 [0.004]

0.01 [0.01]

present

present

Ammodramus savannarum

0.00 [0.00]

0.01 [0.01]

present

present

Eremophila alpestris

undetected

undetected

present

present

Circus cyaneus

present

undetected

present

present

Ring-necked Pheasant

Phasianus colchicus

0.03 [0.02]

0.13 [0.08]

present

present

Savannah Sparrow

Passerculus sandwichensis

0.37 [0.12]

0.94 [0.32]

present

present

Cistothorus platensis

0.00 [0.00]

0.06 [0.03]

present

present

Pooecetes gramineus

0.01 [0.01]

0.02 [0.01]

present

present

Grassland Specialist Species
Bobolink

A

Dickcissel

AB

Eastern Meadowlark
Grasshopper Sparrow
Horned Lark
Northern Harrier

Sedge Wren

A

Vesper Sparrow

A

AB

101

Table 2.9. Continued.
Common Name

Scientific Name

2005

2006

1991

1992

Other Species
American Bittern

Botaurus lentiginosus

undetected

undetected

present

undetected

American Crow

Corvus brachyrhynchos

undetected

undetected

undetected

present

American Goldfinch

Carduelis tristis

0.08 [0.03]

0.26 [0.07]

present

present

American Robin

Turdus migratorius

present

present

present

undetected

Baltimore Oriole

Icterus galbula

undetected

undetected

undetected

present

Blue Jay

Cyanocitta cristata

undetected

undetected

undetected

present

Blue-winged Teal

Anas discors

undetected

undetected

undetected

present

Barn Swallow

Hirundo rustica

NA*

present

undetected

undetected

Brown Thrasher

Toxostoma rufum

undetected

present

undetected

present

Brown-headed Cowbird

Molothrus ater

undetected

present

undetected

undetected

Chipping Sparrow

Spizella passerina

undetected

present

undetected

present

Clay-colored Sparrow

Spizella pallida

present

present

undetected

undetected

Common Yellowthroat

Geothlypis trichas

0.05 [0.04]

0.10 [0.06]

present

present

102

Table 2.9. Continued.
Common Name

Scientific Name

2005

Eastern Bluebird

Sialia sialis

undetected

undetected

undetected

present

Eastern Kingbird

Tyrannus tyrannus

present

0.004 [0.003]

present

present

Eastern Phoebe

Sayornis phoebe

undetected

undetected

undetected

present

European Starling

Sturnus vulgaris

undetected

undetected

present

undetected

Field Sparrow

Spizella pusilla

undetected

undetected

present

present

House Wren

Troglodytes aedon

present

undetected

undetected

present

Indigo Bunting

Passerina cyanea

present

present

present

undetected

Mallard

Anas platyrhynchos

undetected

undetected

present

present

Mourning Dove

Zenaida macroura

present

present

present

undetected

Northern Bobwhite Quail

Colinus virginianus

present

present

undetected

present

Northern Mockingbird

Mimus polyglottos

undetected

undetected

undetected

present

Red-tailed Hawk

Buteo jamaicensis

undetected

undetected

undetected

present

Red-winged Blackbird

Agelaius phoeniceus

2.30 [0.29]

2.97 [0.36]

present

present

Ruby-throated Hummingbird

Archilochus colubris

undetected

undetected

present

present

Song Sparrow

Melospiza melodia

0.46 [0.08]

0.76 [0.14]

present

present

103

2006

1991

1992

Table 2.9. Continued.
Common Name

Scientific Name

2005

Tree Swallow

Tachycineta bicolor

NA*

present

undetected

present

White-throated Sparrow

Zonotrichia leucophrys

undetected

undetected

undetected

present

Yellow Warbler

Dendroica petechia

undetected

undetected

undetected

present

A

Michigan Priority Landbird (Knutson et al. 2001, Matteson et al. 2009)
Michigan Species of Special Concern
* Barn Swallow and Tree Swallow were not included in the 2005 survey.
B

104

2006

1991

1992

CHAPTER 2 LITERATURE CITED

105

Albert, D. A. 1995. Regional landscape ecosystems of Michigan, Minnesota, and
Wisconsin: a working map and classification. Gen. Tech. Rep. NC-178. St. Paul,
MN: U.S. Department of Agriculture, Forest Service, North Central Forest
Experiment Station.

Arenz, C. L., and A. Joern. 1996. Prairie legacies - invertebrates. Pages 91-109 in F. B.
Samson and F. L. Knopf, editors. Prairie conservation: preserving North
America‘s most endangered ecosystem. Island Press, Covelo, California, USA.

BBAE (Breeding Bird Atlas Explorer online resource). 2010. U.S. Geological Survey
Patuxent Wildlife Research Center & National Biological Information
Infrastructure. http://www.pwrc.usgs.gov/bba. Accessed November 2010.

Benedict, R. A., P.W. Freeman, and H. H. Genoways. 1996. Prairie legacies - mammals.
Pages 149-166 in F. B. Samson and F. L. Knopf, editors. Prairie conservation:
preserving North America‘s most endangered ecosystem. Island Press, Covelo,
California, USA.

Brennan, L. A., and W. P. Kuvlesky, Jr. 2005. North American grassland birds: an
unfolding conservation crisis? Journal of Wildlife Management 69:1-13

Bock, C. E., and Z. F. Jones. 2004. Avian habitat evaluation: should counting birds
count? Frontiers in Ecology and the Environment 2:403-410.

Briggs, J. M., A. K. Knapp, J. M. Blair, J. L Heisler, G. A. Hoch, M. S. Lett, and J. K.
McCarron. 2005. An ecosystem in transition: causes and consequences of the
conversion of mesic grassland to shrubland. BioScience 55:243-254.

Burnham, K. P., and D. R. Anderson. 2002. Model Selection and Multi-model
Inference, 2nd Edition. Springer-Verlag, New York, New York, USA.
Corn, P. S., and C. R. Peterson. 1996. Prairie legacies – amphibians and reptiles. Pages
125-134 in F. B. Samson and F. L. Knopf, editors. Prairie conservation:
preserving North America‘s most endangered ecosystem. Island Press, Covelo,
California, USA.

106

Daubenmire, R. F. 1959. A canopy cover method of vegetational analysis. Northwest
Science 33:43-64.

Edwards, D. K., G. L. Dorsey, and J. A. Crawford. 1981. A comparison of three avian
census methods. Studies in Avian Biology 6:170-176.

Fargione, J. E., T. R. Cooper, D. J. Flaspohler, J. Hill, C. Lehman, T. McCoy, S.
McLeod, E. J. Nelson, K. S. Oberhauser, and D. Tilman. 2009. Bioenergy and
wildlife: threats and opportunities for grassland conservation. BioScience 59:767777.

Giudice, J. H., and J. T. Ratti. 2001. Ring-necked Pheasant (Phasianus colchicus), The
Birds of North America Online (A. Poole, Ed.). Cornell Lab of Ornithology,
Ithica, New York, USA. http://bna.birds.cornell.edu/bna/species/176. Accessed
September 2010.

Giuliano, W. M., and S. E. Daves. 2002. Avian response to warm-season grass use in
pasture and hayfield management. Biological Conservation 106:1-9.

Harmeson, J. 1974. Breeding ecology of the Dickcissel. Auk 91:348-359.

Heard, L. P., A. W. Allen, L. B. Best, S. J. Brady, W. Burger, A. J. Esser, E. Hackett, D.
H. Johnson, R. L. Pederson, R. E. Reynolds, C. Rewa, M. R. Ryan, R. T. Molleur,
and P. Buck. 2000. A comprehensive review of Farm Bill contributions to
wildlife conservation, 1985-2000. W. L. Hohman, and D. J. Halloum, editors.
U.S.D.A., Natural Resources Conservation Service, Wildlife Habitat Management
Institute, Technical Report USDA/NRCS/WHMI-2000.

Herkert, J. R. 1995. An analysis of Midwestern breeding bird population trends: 19661993. American Midland Naturalist 134:41-50.

Herkert, J. R., D. E. Kroodsma, and J. P. Gibbs. 2001. Sedge Wren (Cistothorus
platensis), The Birds of North America Online (A. Poole, Ed.). Cornell Lab of
Ornithology, Ithica, New York, USA. http://bna.birds.cornell.edu/bna/
species/176. Accessed September 2010.

107

Herkert, J. R. 2009. Response of bird populations to farmland set-aside programs.
Conservation Biology 23:1036-1040.

Hoekstra, J. M., T. M. Boucher, T. H. Ricketts, and C. Roberts. 2005. Confronting a
biome crisis: global disparities of habitat loss and protection. Ecology Letters
9:23-29.
Hines, J. E. 2006. PRESENCE 2 – Software to Estimate Patch Occupancy and Related
Parameters. United States Geological Survey – Patuxent Wildlife Research
Center, Laurel, Maryland, USA. http://www.mbr-pwrc.usgs.gov/
software/presence.html. Accessed May 2009.

Jones, S. L., and J. E. Cornely. 2002. Vesper Sparrow (Pooecetes gramineus), The Birds
of North America Online (A. Poole, Ed.). Cornell Lab of Ornithology, Ithica,
New York, USA. http://bna.birds.cornell.edu/bna/species/176. Accessed
September 2010.

Knopf, F. L. 1994. Avian assemblages on altered grasslands. Studies in Avian Biology
15:247-257.

Knopf, F. L. 1996. Prairie legacies - birds. Pages 115-141 in F. B. Samson and F. L.
Knopf, editors. Prairie conservation: preserving North America‘s most
endangered ecosystem. Island Press, Covelo, California, USA.

Knutson, M. G., G. Butcher, J. Fitzgerald, and J. Shieldcastle. 2001. Partners in Flight
Bird Conservation Plan for The Upper Great Lakes Plain (Physiographic Area
16). USGS Upper Midwest Environmental Sciences Center in cooperation with
Partners in Flight. La Crosse, Wisconsin, USA.

Lanyon, W. E. 1995. Eastern Meadowlark (Sturnella magna), The Birds of North
America Online (A. Poole, Ed.). Cornell Lab of Ornithology, Ithica, New York,
USA. http://bna.birds.cornell.edu/bna/species/176. Accessed September 2010.

Leach, M. K., and T. J. Givnish. 1996. Ecological determinants of species loss in
remnant prairies. Science 273:1555-1558.

108

MacKenzie, D. I., J. D. Nichols, G. B. Lachman, S. Droege, J. A. Royle, and C. A.
Langtimm. 2002. Estimating site occupancy rates when detection probabilities
are less than one. Ecology 83:2248–2255.

MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines.
2006. Occupancy Estimation and Modeling: Inferring Patterns and Dynamics of
Species Occurrence. Academic Press, Burlington, Massachusetts, USA.

MacKenzie, D. I., J. D. Nichols, M. E. Seamans, and R. J. Gutiérrez. 2009. Modeling
species occurrence dynamics with multiple states and imperfect detection.
Ecology 90:823-835.

Martin, S. G., and T. A. Gavin. 1995. Bobolink (Dolichonyx oryzivorus), The Birds of
North America Online (A. Poole, Ed.). Cornell Lab of Ornithology, Ithica, New
York, USA. http://bna.birds.cornell.edu/bna/species/176. Accessed September
2010.

Matteson, S., K. Kreitinger, G. Bartelt, G. Butcher, D. Sample, and T. Will. 2009.
Partners in Flight Bird Conservation Plan for The Boreal Hardwood Transition
(Bird Conservation Region 12 — U.S. Portion). Version 1.0. Partners in Flight.
www.partnersinflight.org. Accessed November 2010.

McCoy, T. D., M. R. Ryan, and L. W. Burger. 2001. Grassland bird conservation: CP1
vs. CP2 plantings in Conservation Reserve Program fields in Missouri. American
Midland Naturalist 145:1-17.

Millenbah, K. F. 1993. The effects of different age classes of fields enrolled in the
Conservation Reserve Program in Michigan on avian diversity, density, and
productivity. M.S. Thesis, Michigan State University, East Lansing, Michigan,
U.S.A.

Murray, L. D., C. A. Ribic, and W. E. Thogmartin. 2008. Relationship of obligate
grassland birds to landscape structure in Wisconsin. Journal of Wildlife
Management 72:463-467.

109

Nichols, J. D., L. L. Bailey, A. F. O‘Connell, Jr., N. W. Talancy, E. H. Campbell Grant,
A. T. Gilbert, E. M. Annand, T. P. Husband, and J. E. Hines. 2008. Multi-scale
occupancy estimation and modeling using multiple detection methods. Journal of
Applied Ecology 45:1321-1329.

Norment, C. J., C. D. Ardizzone, and K. Hartman. 1999. Habitat relations and birding
biology of grassland birds in western New York. Studies in Avian Biology
19:112-121.

Noss, R. F., E. T. Laroe, and J. M. Scott. 1995. Endangered ecosystems of the United
States: a preliminary assessment of loss and degradation. National Biological
Service, U.S. Department of the Interior, Report No. 0611-R-01 (MF).
Washington, D. C., USA.

Perot, A., and M.-A. Villard. 2009. Putting density back into the habitat-quality
equation: case study of an open-nesting forest bird. Conservation Biology
23:1550-1557.
Rabeni, C. F. 1996. Prairie legacies – fish and aquatic resources. Pages 111-124 in F. B.
Samson and F. L. Knopf, editors. Prairie conservation: preserving North
America‘s most endangered ecosystem. Island Press, Covelo, California, USA.

Ribic, C.A., R. R. Koford, J. R. Herkert, D. H. Johnson, N. D. Niemuth, D. E. Naugel, K.
K. Bakker, D. W. Sample, and R. B. Renfrew. 2009. Area sensitivity in North
American grassland birds: patterns and processes. The Auk 126:233-244.

Riffell, S., D. Scognamillo, and L.W. Burger. 2008. Effects of the Conservation Reserve
Program on northern bobwhite and grassland birds. Environmental Monitoring
and Assessment 146:309-323.

Sampson, F. B., and F. L. Knopf, editors. 1996. Prairie conservation: preserving North
America‘s most endangered ecosystem. Island Press, Washington, D.C., USA.
Sargent, M. and K. S. Carter, editors. 1999. Managing Michigan‘s wildlife: a
landowner‘s guide. Michigan United Conservation Clubs, East Lansing,
Michigan, USA.

110

Sauer, J. R., B. G. Peterjohn, S. Schwartz, and J. E. Hines. 1995. The grassland bird
home page. Version 95.0. Patuxent Wildlife Research Center, Laurel, Maryland,
USA. http://www.mbr-pwrc.usgs.gov/bbs/grass/grass.htm. Accessed May 2010.

Sauer, J. R., J. E. Hines, and J. Fallon. 2008. The North American Breeding Bird
Survey, Results and Analysis 1966 - 2007. Version 5.15.2008. USGS Patuxent
Wildlife Research Center, Laurel, Maryland, USA. http://www.mbrpwrc.usgs.gov/bbs/bbs.html. Accessed November 2010.

Searchinger T., R. Heimlich, R. A. Houghton, F. X. Dong, A. Elobeid, J. Fabiosa, S.
Tokgoz, D. Hayes, and T. H. Yu. 2008. Use of US croplands for biofuels
increases greenhouse gases through emissions from land-use change. Science
319: 1238–1240.

Secchi S. and B. A. Babcock. 2007. Impact of high corn prices on Conservation Reserve
Program acreage. Iowa Ag Review (Spring): 13.

Suter, J. F., G. L. Poe, and N. L. Bills. 2008. Do landowners respond to land retirement
incentives? Evidence from the Conservation Reserve Enhancement Program.
Land Economics 84:17-30.

Temple, S. A. 2002. Dickcissel (Spiza americana), The Birds of North America Online
(A. Poole, Ed.). Cornell Lab of Ornithology, Ithica, New York, USA.
http://bna.birds.cornell.edu/bna/species/176. Accessed September 2010.

USDA (U.S. Department of Agriculture). 2000a. Introduced grass planting:
Conservation Reserve Enhancement Program, CREP CP-1. Natural Resources
Conservation Service, Michigan. http://www.michigan.gov/documents/mda_ CP1_INTRODUCEDGRASS_9614_7.pdf. Accessed May 2010.

USDA (U.S. Department of Agriculture). 2000b. Native grass planting: Conservation
Reserve Enhancement Program, CREP CP-2. Natural Resources Conservation
Service, Michigan. http://www.michigan.gov/documents/mda_CP-2_
NATIVEGRASS_9615_7.pdf. Accessed May 2010.

USDA (U.S. Department of Agriculture). 2007. Conservation Reserve Program
summary and enrollment statistics, fiscal year 2006. Farm Service Agency.
http://www.fsa.usda.gov/ Internet/FSA_File/06rpt.pdf. Accessed May 2010.
111

USDA NRCS (U.S. Department of Agriculture Natural Resources Conservation Service).
2011. The PLANTS Database. National Plant Data Center, Baton Rouge,
Louisiana, USA. http://plants.usda.gov. Accessed January 2011.

Van Horne, B. 1983. Density as a misleading indicator of habitat quality. Journal of
Wildlife Management 47:893-901.

Veech, J.A. 2006. A comparison of landscapes occupied by increasing and decreasing
populations of grassland birds. Conservation Biology 20:1422-1432.

Vickery, P. D. 1996. Grasshopper Sparrow (Ammodramus savannarum), The Birds of
North America Online (A. Poole, Ed.). Cornell Lab of Ornithology, Ithica, New
York, USA. http://bna.birds.cornell.edu/bna/species/176. Accessed September
2010.

Voss, E. G. 1985. Michigan Flora: a guide to the identification and occurrence of the
native and naturalized seed-plants of the state; Part II: Dicots (SaururaceaeCornaceae). Cranbrook Institute of Science Bulletin 59 and University of
Michigan Herbarium, Ann Arbor, Michigan, USA.

Voss, E. G. 1996. Michigan Flora: a guide to the identification and occurrence of the
native and naturalized seed-plants of the state; Part III: Dicots (PyrolaceaeCompositae). Cranbrook Institute of Science Bulletin 61 and University of
Michigan Herbarium, Ann Arbor, Michigan, USA.

Walk, J. W., E. L. Kershner, T. J. Benson, and R. E. Warner. 2010. Nesting success of
grassland birds in small patches in an agricultural landscape. The Auk 127:328334.

Wentworth, K. L., M. C. Brittingham, and A. M. Wilson. 2010. Conservation Reserve
Enhancement Program fields: benefits for grassland and shrub-scrub species.
Journal of Soil and Water Conservation 65:50-60.

Wheelwright, N. T., and J. D. Rising. 2008. Savannah Sparrow (Passerculus
sandwichensis), The Birds of North America Online (A. Poole, Ed.). Cornell Lab
of Ornithology, Ithica, New York, USA. http://bna.birds.cornell.edu/bna/species/
176. Accessed September 2010.

112

CHAPTER 3: GRASSLAND BIRD DETECTION AND OCCUPANCY IN
MICHIGAN CONSERVATION RESERVE ENHANCEMENT PROGRAM
(CREP) GRASSLANDS

INTRODUCTION
Breeding Bird Survey (BBS) data indicate that grassland birds have declined
more rapidly, more consistently, and over a wider geographic area than any other guild of
North American birds (Sauer et al. 1995, Knopf 1996). Six grassland bird species found
in Michigan are among those that have been designated as Michigan priority land birds
by the bird conservation group Partners in Flight: Bobolink, Dickcissel, Grasshopper
Sparrow, Henslow‘s Sparrow, Northern Harrier, and Sedge Wren (Knutson et al. 2001,
Matteson et al. 2009). Dickcissel, Grasshopper Sparrow, and Northern Harrier are also
state species of special concern, and Henslow‘s Sparrow is a state endangered species.
Native temperate grasslands currently have the lowest protected area compared to
converted area of any major biome (Hoekstra et al. 2005) and are one of the most
endangered ecosystems in the world (Sampson and Knopf 1996). Farmland set-aside
programs such as the Conservation Reserve Enhancement Program (CREP) may diminish
or reverse the decline of grassland bird species by increasing the area of native and nonnative grasslands in the U.S. (Herkert 2009). Geographic variation in the effects of setaside lands on wildlife exists due to local and regional climatic differences and variation
in program implementation, including differences in management practices (i.e.,
prescribed burning; Heard et al. 2000, Riffell et al. 2008). As a result, it is important to
conduct assessments of the impacts of farmland set-aside programs on grassland birds

113

and other wildlife at sub-national scales, such as state, ecoregional, or major watershed
levels.
Grassland bird habitat requirements have also been found to vary among
geographic regions (e.g., Johnson and Igl 2001, Winter et al. 2006). For example, Winter
et al. (2006) found that the density of 3 grassland bird species varied across relatively
small regional scales (i.e., separated by about 50 km) in the tallgrass prairie of
southeastern North Dakota and northwestern Minnesota. Regional differences in
grassland bird habitat selection highlight the importance of evaluating grassland bird
habitat use at local scales, consistent with this assessment of grassland bird occupancy in
CREP fields of the Saginaw Bay watershed in Michigan. This is the first study in
Michigan to examine grassland bird occupancy of CREP grasslands. The goal of this
analysis was to model grassland bird occupancy as a function of CREP field size and type
for all observed grassland bird species that were adequately detected within two different
types of CREP grassland plantings, CREP conservation practice (CP) 23 and CP1 fields.
METHODS
2

This study occurred within the 8,219 km Sandusky Lake Plain sub-subsection of
Michigan‘s Lower Peninsula (Albert 1995). This sub-subsection consists of flat clay lake
plain, bordered by shoreline dunes and sand plain. In the center of the flat clay lake
plain, long and narrow till plains and end moraine ridges parallel either Saginaw Bay or
Lake Huron (Albert 1995). Prior to European settlement, extensive wet and wet-mesic
prairies occurred inland from the coastal Great Lakes marshes of Saginaw Bay, but
largely due to agricultural conversion, these prairies now occur only as small remnants,
generally on State-owned lands (Albert 1995).

114

In Michigan, 22,983 ha of privately-owned land in the Saginaw Bay, Lake
Macatawa, and River Raisin watersheds were enrolled in CREP at the time of this study
(USDA 2007). From CREP lands in the Saginaw Bay watershed, 14 CREP grassland
fields in Tuscola County, ranging in size from 6.8 to 23.6 ha, were selected for study
(Figure 3.1). All grassland study fields were planted in 2002 and were surveyed for birds
in 2005 and 2006. Seven study fields were CREP conservation practice (CP) type 23
(hereafter CP23) that had been planted with a seed mixture of native grasses and native
and introduced wildflowers over 100% of the field area, and 7 study fields were CP type
1 (hereafter CP1) fields that had been planted with introduced grasses and legumes over
70% of the field area, with the remaining 30% of the field area planted with the same
native seed mixture used in CP23 fields (Figure 3.1). Native sections of CP1 fields were
established to provide winter shelter for wildlife (e.g., Ring-necked Pheasant, Phasianus
colchicus) because native grasses tend to remain upright under winter conditions whereas
introduced grasses generally collapse (USDA 2000a, USDA 2000b). In two CP1 study
fields, the native seed mixture did not successfully establish and the field area was
composed only of introduced vegetation (Figure 3.1). In 2005, the native seed mixture
cost $16.00/pound (i.e., $104.00/acre or $256.99/hectare) and the introduced seed
mixture cost $1.90/pound (i.e., $20.90/acre or $51.65/hectare) plus 6% sales tax through
the Tuscola County Conservation District.
Within CREP, CP23 fields are wetland restorations and all CP23 fields had hydric
soils. A small wetland (i.e., 3-10% of upland area) was created in each CP23 field by
cutting drain tiles, and if necessary, by creating a push-out, or shallow depression in the
soil. The CREP program offers a whole-field native grassland conservation practice

115

without a wetland component, CP2, which was rare in the Saginaw Bay area. Due to the
scarcity of CP2 fields, the upland portions of CP23 fields were used for comparison to the
introduced portions of CP1 fields and to CP1 whole fields. Wetlands and a buffer of 25
m surrounding the wetlands were not included in CP23 avian surveys. The surveyed (i.e.,
upland) area of CP23 fields ranged in size from 6.9 – 19.8 ha (median [SE] = 12.6 [1.9]
ha, median = 11.8 ha). CP1 whole fields ranged in size from 7.3 – 23.9 ha (mean [SE] =
12.3 [2.1] ha, median = 10.9 ha), and the introduced portion of CP1 fields ranged in size
from 5.5 – 17.0 ha (mean [SE] = 9.4 [1.5], median = 8.0 ha).
Avian data were collected using line-transect sampling (Edwards et al. 1981).
The total area of each study field (i.e., except CP23 wetlands and associated wetland
buffers) was surveyed every 2 weeks from 1 June – 14 August in 2005 (n=5 surveys in
each field) and from 15 May – 14 August in 2006 (n=6 surveys per field). Surveys took
place along transects running the length of each field. The first transect was established
25 m from a randomly selected field corner and subsequent transects were established at
50 m intervals across the width of the field. An investigator recorded the species and
location of each individual bird observed. If an individual bird was flushed by the
observer, the new location of the bird was noted in order to prevent counting the same
bird more than once. If a bird was observed but could not be identified, the location of
the bird was recorded and the species was designated as unknown. Unidentified
individuals were not included in analyses. Avian surveys were conducted after sunrise
and before 12:00 p.m., and surveys were not performed in the rain or if wind speeds were
over 16 kph.

116

Avian species detection, p, and occupancy, ψ, probabilities in CP23 and CP1
study fields were estimated using single-season occupancy models (MacKenzie et
al.2002) in the program PRESENCE (Hines 2006). The use of single season models was
based on the results of likelihood ratio testing (MacKenzie et al. 2006, Nichols et al.
2008), which indicated that the more highly parameterized multi-season models did not
fit the data significantly better than single season models (p > 0.05).
For studies with small sample sizes, model ranking using Akaike‘s information
criterion corrected for small sample sizes (AICc) is recommended (Burnham and
Anderson 2002). The top-ranking detection model for each avian species was identified
as having the lowest AICc value among all candidate models with constant occupancy,
ψ(·) (MacKenzie et al. 2002). Candidate (i.e., successful) detection models were those
that achieved 1) numerical convergence of parameter estimates to >6 significant figures,
2) a balanced variance-covariance matrix, and 3) small- or moderately-sized standard
errors of betas, the untransformed estimates of coefficients for covariates (i.e., standard
error was no greater than 5 times the corresponding beta) (D. MacKenzie, Program
PRESENCE – FAQ at http://www.phidot.org/forum, accessed October 2009).
Avian detection probability was calculated for each species observed in at least
10% of the total number of surveys (MacKenzie et al. 2002). Detection probability
estimates were obtained from the top-ranking detection model for each species. Potential
detection covariates were ordinal date and CREP field type. Ordinal date was
standardized to reduce biases in parameter estimation (MacKenzie et al. 2002). When
field type and/or ordinal date were covariates in the top-ranking detection model,
multiple estimates of detection resulted, and the lowest estimate was reported. For
117

species where no detection models were successfully generated using the full data set,
detection probability estimates for individual study sites were examined, and if detection
probability estimates for a subset of the data appeared adequate (e.g., detection estimates
for all fields in 2006 were ≥0.10), then detection modeling was performed on this data
subset. For species where the naïve estimate of ψ equaled 1, study fields were fully
occupied, and detection modeling was conducted with ψ fixed to 1.
Occupancy modeling was performed only for species with detection probability
estimates ≥0.10. Potential occupancy covariates were ordinal date, CREP field type, and
CREP field size. Ordinal date and field size were standardized to reduce biases in
parameter estimation (MacKenzie et al. 2002).
For each avian species with a minimum detection probability estimate ≥ 0.10, all
possible occupancy models were constructed using the best-fitting detection model as a
function of the occupancy covariates (i.e., ψ[covariates], p[as in best fitting detection
model]) (Roloff et al. in press). Candidate occupancy models met the convergence,
variance-covariance matrix, and standard error size criteria as described above. The
global model was identified as the most parameterized occupancy model. Global model
fit was evaluated using the chi-square goodness-of-fit statistic to estimate an
overdispersion parameter, ĉ, from 2,000 bootstrap iterations (MacKenzie and Bailey
2004, MacKenzie et al. 2006). Ecological models are commonly overdispersed (i.e., ĉ ≠
1), and adjustment of the model selection criteria to account for it is advised (MacKenzie
and Bailey 2004, MacKenzie et al. 2006). Overdispersed occupancy models were ranked
using QAICc, an adjusted AICc with standard errors inflated using ĉ, the overdispersion
parameter (McCullagh and Nelder 1989, MacKenzie et al. 2006). High overdispersion
118

parameter values of a global model indicate that any derived occupancy models would
not have useful structure (Burnham and Anderson 1998), so for species with extremely
overdispersed global models (i.e., ĉ > 4), no occupancy models were successful. For
species with suitable global models (i.e., ĉ ≤ 4), all occupancy models in the set of
candidate models were ranked according to AICc or QAICc (Burnham and Anderson
2002). When one best occupancy model did not emerge, competing models (i.e., models
where ∆AICc or ∆QAICc ≤ 2.00) were model averaged to derive a composite model
(Burnham and Anderson 2002).
RESULTS
A total of 34 avian species were observed in CREP study fields during the 2 years
of this study (Appendix 3.2). In 2005, 23% of individual birds observed during surveys
were unidentified, and in 2006 surveyors were more experienced and only 8% of
observed individuals were unidentified. Using the grassland bird list developed for the
BBS by the U.S. Geographic Survey (Sauer et al. 1995), 11 of the observed species were
identified as grassland specialists: Bobolink (Dolichonyx oryzivorus), Dickcissel (Spiza
americana), Eastern Meadowlark (Sturnella magna), Grasshopper Sparrow
(Ammodramus savannarum), Henslow‘s Sparrow (Ammodramus henslowii), Horned
Lark (Eremophila alpestris), Northern Harrier (Circus cyaneus), Ring-necked Pheasant,
Savannah Sparrow (Passerculus sandwichensis), Sedge Wren (Cistothorus platensis),
and Vesper Sparrow (Pooecetes gramineus). Six of these grassland specialist species are
of special conservation priority in Michigan (Appendix 3.2). Bobolink, Dickcissel,
Grasshopper Sparrow, Henslow‘s Sparrow, Northern Harrier, and Sedge Wren are
designated as Michigan priority landbirds by the bird conservation group Partners in
119

Flight (Knutson et al. 2001, Matteson et al. 2009), while Dickcissel, Grasshopper
Sparrow, and Northern Harrier are also state species of special concern and Henslow‘s
Sparrow is a state endangered species (Appendix 3.2).
Bobolink and Savannah Sparrow were the most abundant grassland bird species
in native and introduced grasslands (Appendix 3.2). Three grassland bird species were
too rarely observed to estimate their detection probabilities (i.e., observed in less than
10% of surveys): Henslow‘s Sparrow, Horned Lark, and Northern Harrier. Detection
probability modeling was performed for the 8 other grassland specialist species observed
during this study. Likelihood ratio testing indicated that more highly parameterized
multi-season models did not fit the data significantly better than simpler single-season
models (p > 0.05), so single season models were used in all detection and occupancy
modeling efforts (MacKenzie et al. 2006, Nichols et al. 2008).
Adequate detection probability estimates (i.e., p ≥ 0.10) were achieved for all 8
species, with detection probability estimates ranging from 0.12 to 0.81 (Table 3.1).
However, an adequate detection probability estimate was obtained only when a subset of
the data was used in the case of one species, Sedge Wren (2006 data), most likely
because Sedge Wren were rare in 2005 (mean density = 0.01, SE = 0.01 in CP23 fields;
mean density = 0.0, SE = 0.0 in introduced vegetation of CP1 fields; 2005 Sedge Wren
mean density information was reported for CP23 fields and CP1 introduced vegetation in
Appendix 3.2 because detection probability estimates were adequate for Sedge Wren in
native and introduced CREP grassland plant communities [Chapter 4]).
CREP field type and/or standardized ordinal date were included in the best
detection models for Bobolink, Eastern Meadowlark, Ring-necked Pheasant, Savannah

120

Sparrow, and Sedge Wren (Table 3.1). The detection probability of Bobolink varied
between 2005 and 2006, but detection probability estimates were >0.63 regardless of
year, indicating that Bobolink was readily detected throughout each breeding season
(Figure 3.2). Eastern Meadowlark was reliably detected in CP23 and CP1 fields, but the
species tended to be more likely to be detected in CP23 fields, although the 95%
confidence intervals for detection estimates of the two field types were overlapping (i.e.,
95% confidence interval was 0.10 – 0.16 in CP23 fields and 0.08 – 0.15 in CP1 fields;
Figure 3.3). Ring-necked Pheasant detection probability varied between 2005 and 2006,
but detection probability estimates were >0.17 in both years, indicating the species was
reliably detected throughout both breeding seasons (Figure 3.4). Ring-necked Pheasant
was adequately detected in CP23 and CP1 fields, but the species tended to be more likely
to be detected in CP23 fields, although the 95% confidence intervals for detection
estimates of the two field types were overlapping (i.e., 95% confidence interval was 0.21
– 0.46 in CP23 fields and 0.05 – 0.32 in CP1 fields; Figure 3.4). Detection probability
estimates for Savannah Sparrow were >0.81 regardless of year, indicating the species was
readily detected throughout each breeding season (Figure 3.5). The detection probability
of Sedge Wren was >0.15 throughout 2006, indicating the species was reliably detected
throughout the breeding season, but Sedge Wren detection probability was positively
associated with standardized ordinal date, suggesting that optimal detection of this
species occurred toward the end of the breeding season (Figure 3.6).
Occupancy modeling was performed for all 8 grassland bird species with
detection probabilities ≥0.10 (i.e., Bobolink, Dickcissel, Eastern Meadowlark,
Grasshopper Sparrow, Ring-necked Pheasant, Savannah Sparrow, Sedge Wren, and

121

Vesper Sparrow; Table 3.1). The naïve estimate of occupancy was 1.0 for Ring-necked
Pheasant and Savannah Sparrow, indicating that all surveyed fields were occupied.
Global occupancy model fit was evaluated for the remaining 6 grassland bird species, and
for 2 species, Bobolink and Eastern Meadowlark, the global occupancy models were
extremely overdispersed (i.e., ĉ > 4; Table 3.2), indicating the occupancies of these 2
species could not be reliably estimated from this study. Global model fit indicated that
informative occupancy models could be derived for the 4 remaining grassland bird
species: Dickcissel, Grasshopper Sparrow, Sedge Wren (2006 data), and Vesper Sparrow
(Table 3.2).
A single best-fitting occupancy model was identified for Dickcissel that included
field size and was not overdispersed (ĉ = 0.90) (Table 3.3). This model accounted for
64% of the total AICc weight among four candidate models (Table 3.3). The naïve
estimate of occupancy was 0.43, while the detection corrected estimates of occupancy by
site ranged from 0.003 to 0.85. The beta estimate for the effect of standardized field size
on occupancy was negative (beta = -2.21, SE = 1.44; Table 3.4), suggesting that larger
CREP fields had lower Dickcissel occupancy. However, the 95% confidence interval of
the beta estimate indicated that field size was not a significant determinant of Dickcissel
occupancy (Table 3.4).
For Grasshopper Sparrow, the intercept-only model ranked highest (ĉ = 2.79; w =
0.84; Table 3.5), indicating that none of the covariates tested in this study were important
correlates of Grasshopper Sparrow occupancy. The naïve estimate of occupancy was
0.64, and the detection corrected occupancy estimate was 0.65.

122

Occupancy modeling for Sedge Wren (2006 data) resulted in two competing
models (i.e., ∆AICc ≤ 2) that together accounted for 80% of the total AICc weight (Table
3.6). The highest ranking occupancy model was the intercept-only model (ĉ = 1.65), and
the competing model included CREP field type (ĉ = 1.61; Table 3.6). The naïve
occupancy estimate for Sedge Wren was 0.64, and the detection corrected occupancy
estimates by site ranged from 0.44 in CP1 fields to 0.88 in CP23 fields. The model
averaged beta estimate of the effect of CREP field type on occupancy was negative (beta
= -2.25, SE = 1.53; Table 3.4), indicating that Sedge Wren occupancy was greater in
CP23 fields than in CP1 fields (i.e., CP23 fields were assigned a value of 0 and CP1
fields were categorized as 1 during modeling). However, the 95% confidence interval for
the field type parameter included zero indicating that CREP field type was not a
significant determinant of Sedge Wren occupancy (Table 3.4).
Two competing occupancy models (i.e., ∆AICc ≤ 2) were identified for Vesper
Sparrow that together accounted for 87% of the total AICc weight among 3 candidate
models (Table 3.7). Similar to the findings for Sedge Wren, the intercept-only model (ĉ
= 0.70) was the top-ranked occupancy model for Vesper Sparrow and the competing
model included CREP field type (ĉ = 0.67; Table 3.7). The naïve estimate of occupancy
was 0.86, while the detection corrected estimates of occupancy by site ranged from 0.76
in CP1 fields to 1.0 in CP23 fields. The beta estimate for CREP field type on occupancy
was negative (beta = -21.83, SE = 13.42; Table 3.4), indicating that Sedge Wren
occupancy was greater in CP23 fields than in CP1 fields (i.e., CP23 fields were assigned
a value of 0 and CP1 fields were categorized as 1 during modeling). However, the 95%

123

confidence interval of the field type beta estimate included zero indicating that CREP
field type was not a significant determinant of Vesper Sparrow occupancy (Table 3.4).
DISCUSSION
Avian surveys of CP23 and CP1 CREP fields in Michigan‘s Saginaw Bay
watershed resulted in the observation of 34 species, 11 of which were grassland specialist
species (Sauer et al. 1995). Sufficient numbers of 8 grassland species were observed to
allow detection probability and occupancy modeling (i.e., present in ≥10% of surveys).
All 8 species had adequate detection probability estimates (i.e., ≥ 0.10), which were
above 0.25 for 4 species, indicating that the survey technique used in this study was an
appropriate methodology for documenting occupancy. Covariates included in the topranking occupancy models of grassland bird species included field type (i.e., CP23 or
CP1) and CREP field size. Although the effects of these covariates were not significant,
findings of this study are likely constrained by small sample sizes, and these covariates
warrant evaluation in further research.
Grassland bird species that occurred in this study were similar to those
documented in comparable research. Published studies on grassland bird species in
CREP fields of the Lake States are rare; only one study in Pennsylvania was identified
(Wentworth et al. 2010). Wentworth et al. (2010) found 8 grassland bird species that
were also observed in my study: Bobolink, Dickcissel, Eastern Meadowlark, Grasshopper
Sparrow, Henslow‘s Sparrow, Ring-necked Pheasant, Savannah Sparrow, and Vesper
Sparrow (i.e., Horned Lark, Northern Harrier, and Sedge Wren were not observed in
Pennsylvania but were present in Michigan CREP grasslands). The grassland bird
species that occurred in my study were also similar to those that occurred in comparable

124

work in grasslands of a similar farmland set-aside program, the Conservation Reserve
Program (CRP). In an investigation of the effects of different age classes of Michigan
CRP fields on avian diversity, density, and productivity, Millenbah (1993) found 10
grassland bird species, all of which were also observed in my study along with Henslow‘s
Sparrow. Nine grassland bird species observed in my surveys also occurred in CRP
fields in a large-scale analysis of Midwestern bird abundance and nesting in CRP fields
and cropland (i.e., Horned Lark and Northern Harrier were absent in CRP fields) (Best et
al. 1997). With the exception of Savannah Sparrow, the grassland species that were
observed in my surveys were also observed in a study of CP1 and switchgrass (Panicum
virgatum) dominated CP2 CRP fields in Missouri (McCoy et al. 2001). These
comparisons indicate that grassland bird community composition is similar in CREP and
CRP, however other research suggests that grassland bird abundances vary within and
between CREP and CRP (Chapter 2, McCoy et al. 2001, Wentworth 2010), underscoring
the importance of evaluating CREP and other such conservation programs at local scales
(e.g., within states, ecoregions, or major watersheds). It is important to note that the
consistency of results from studies in different regions may be affected by differences in
grassland composition, age or successional stage, and structure.
The survey methodology of this study generated adequate detection probability
estimates for all grassland species observed in sufficient numbers to perform detection
probability modeling (i.e., observed in ≥10% of surveys). However, it is possible to
recommend methodological improvements for future studies in this or similar study
systems. Improvements to survey methodology to increase the probability of detection
would include restricting avian survey efforts to the period of the breeding season prior to

125

the latest truncation date of expected species or species of interest. Additionally,
increasing the number of surveys prior to the truncation date would improve the
probability of detecting rare species, and increasing the number of fields surveyed would
ease or remove analytical constraints due to small sample sizes.
With one exception, global occupancy models included only one occupancy
covariate, likely due to constraints caused by small sample size (i.e., n = 14). Despite the
constraints of small sample size, field size or field type appeared in top-ranking
occupancy models for three of the four grassland bird species for which occupancy
modeling was successful. The 95% confidence intervals for field size and field type
consistently included 0, but 0 occurred near the tails of the covariate distributions
suggesting that non-zero central tendencies existed. Hence, future studies should strive to
include more replicates to further evaluate the correlations between field size and field
type on grassland bird occupancy.
I expected a relationship between CREP field type and grassland bird occupancy
because of the different proportions of native plant community area in CP23 (100%) and
CP1 (70%) CREP fields. Due to the coevolutionary history of native grasses and
grassland birds (Sampson and Knopf 1996) and clear differences in native and introduced
plant community composition and structure (Chapter 2), I anticipated that grassland bird
occupancy would be higher in CP23 fields than in CP1 fields. For the species exhibiting
a relationship to field type (i.e., Sedge Wren and Vesper Sparrow), higher occupancy
estimates were consistently associated with CP23 fields. My results indicate that these
100% native fields are providing habitat conditions that support higher occupancy for
some grassland bird specialists.

126

Because other research has shown that habitat patch size is an important positive
influence on grassland bird occupancy (e.g., Ribic et al. 2009), I predicted that larger
CREP field size would be associated with higher occupancies of grassland birds in my
study. CREP field size emerged as a correlate to the occupancy Dickcissel, and
occupancy estimates among sites exhibited a broad range (0.01 to 0.86) in response to
field size. Unexpectedly, field size was negatively related to Dickcissel occupancy, with
higher occupancy estimates for smaller-sized CREP fields. My finding contradicts other
research that found a positive relationship of patch size to Dickcissel occurrence (Bakker
et al. 2002).
Combinations of within-field and landscape-scale variables have been identified
as important covariates in other studies of grassland birds (e.g., Ribic and Sample 2001,
Bakker et al. 2002, Cunningham and Johnson 2006, Winter et al. 2006). Grassland birds
likely use a hierarchical process that is affected by variables at multiple scales when
choosing breeding territories (Ribic et al. 2009, Fisher and Davis 2010). For example,
Winter et al. (2006) found that grassland bird density was affected by a combination of
habitat patch size, landscape, region, and local vegetation structure more than by local
vegetation structure alone. In further analyses of my data, I will model grassland bird
occupancy in native plant communities of CP23 CREP fields and introduced plant
communities of CP1 CREP fields and examine the effect of within-field and landscapescale variables on grassland bird occupancy (Chapter 4).
Management recommendations
My occupancy modeling results indicate that CREP fields, especially CP23 native
grasslands, provide habitat for grassland birds in Michigan, a critical resource for an

127

avian guild whose decline has occurred more rapidly, more consistently, and over a wider
geographic area than any other guild of North American birds (Sauer et al. 1995, Knopf
1996). Naïve occupancy estimates were particularly high for Ring-necked Pheasant and
Savannah Sparrow in all study fields. Occupancy estimates of several other grassland
bird species (i.e., Grasshopper Sparrow, Sedge Wren, and Vesper Sparrow) were higher
in the native CP23 grasslands than in CP1 fields, thus I recommend that CREP
administrators prioritize the establishment of CP23 fields within the Saginaw Bay area of
Michigan.
Because occupancy covariates differed among grassland species, I recommend
that in similar studies the effect of occupancy covariates be evaluated for individual
species within the grassland bird guild rather than for grassland birds as a group. I also
advocate local (e.g., ecoregion, major watershed) evaluation of grassland bird habitat
selection, since grassland bird habitat requirements have been found to vary among
geographic regions (e.g., Johnson and Igl 2001), even at small regional scales, such as
those separated by as little as 50 km (Winter et al. 2006). Additionally, local evaluation
of the impacts of CREP on wildlife conservation are important, since geographic
variation in the effects of set-aside lands on wildlife exists due to local and regional
climatic differences and variation in program implementation, including differences in
management practices (Heard et al. 2000, Riffell et al. 2008).
Maintaining and increasing the area of CREP grasslands, especially native
grasslands, should be a grassland bird conservation strategy in Michigan. A trend toward
reconversion of lands enrolled in set-aside programs such as CREP to crop production
has been identified (Fargione et al. 2009), and I recommend that program administrators

128

provide incentives for landowners to join or continue in CREP through an increase in the
one-time program sign-up bonus when possible, as this strategy that has been found to
generate a greater increase in CREP enrollment than did smaller increases in the annual
payments made over the term of the CREP contract (Suter et al. 2008).

129

APPENDIX 3.1: CHAPTER 3 TABLES AND FIGURES

130

CP1 Field
(standard)
n=5

CP23 Field
n=7

Native
Vegetation

Example
Wetland and
25 m Wetland
Buffer

Introduced
Vegetation
(70% area)

CP1 Field
(native planting failed)
n=2
Native
Vegetation
(30% area)

Introduced
Vegetation

Fig. 3.1. Location, type, and number of Conservation Reserve Enhancement Program
(CREP) study fields within Tuscola County, Saginaw Bay watershed, Michigan, 2005
and 2006. Wetlands and wetland buffers were excluded from avian surveys and
vegetation sampling. The surveyed (i.e., upland) area of CP23 fields ranged in size from
6.9 – 19.8 ha (median [SE] = 12.6 [1.9] ha, median = 11.8 ha). The introduced portion of
CP1 fields ranged in size from 5.5 – 17.0 ha (mean [SE] = 9.4 [1.5], median = 8.0 ha),
and CP1 whole fields ranged in size from 7.3 – 23.9 ha (mean [SE] = 12.3 [2.1] ha,
median = 10.9 ha).

131

Table 3.1. Grassland bird species, with corresponding best-fitting detection models (with
constant occupancy, ψ[·]) and detection probability estimates, observed in CP23 and CP1
Conservation Reserve Enhancement Program (CREP) grasslands, Saginaw Bay
watershed, Michigan, 2005-2006. The lowest detection probability estimate was reported
when multiple detection probability estimates were generated.
Common Name

Detection Model

p(·) [SE]

Bobolink

p(Ordinal Date)

0.63 [0.06]

Dickcissel

p(.)

0.35 [0.06]

Eastern Meadowlark

p(CREP Field Type)

0.13 [0.07]

Grasshopper Sparrow

p(.)

0.31 [0.05]

Ring-necked Pheasant

p(CREP Field Type, Ordinal Date)

0.17 [0.05]

Savannah Sparrow

p(Ordinal Date)

0.81 [0.05]

Sedge Wren, 2006 data

p(Ordinal Date)

0.12 [0.07]

Vesper Sparrow

p(.)

0.21 [0.04]

132

Detection Probability (%)

1.00
0.90
0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00

1-14
June

15-30
June

1-14
July

15-31 1-14
July August

15-31
May

2005

1-14
June

15-30
June

1-14
July

15-31 1-14
July August

2006

Fig. 3.2. Relationship between Bobolink detection probability and ordinal date in CP23 and CP1 Conservation Reserve Enhancement
Program (CREP) grasslands, Saginaw Bay watershed, Michigan, 2005-2006. Error bars indicate 95% confidence intervals.

133

1.00

0.90
Detection Probability (%)

0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
CP23

CP1

CREP Field Type

Fig. 3.3. Relationship between Eastern Meadowlark detection probability and
Conservation Reserve Enhancement Program (CREP) administrative field type in CP23
and CP1 CREP grasslands, Saginaw Bay watershed, Michigan, 2005 and 2006. Error
bars indicate 95% confidence intervals.

134

CP23 Probability of Detection
CP1 Probability of Detection

1.00

Detection Probability (%)

0.90
0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10

0.00
1-14
June

15-30
June

1-14
July

15-31 1-14
July August

15-31
May

1-14
June

15-30
June

1-14
July

15-31 1-14
July August

2006

2005

Fig. 3.4. Relationship between Ring-necked Pheasant detection probability, field type, and ordinal date in CP23 and CP1
Conservation Reserve Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan, 2005 and 2006. Error bars
indicate 95% confidence intervals.

135

1.00
Detection Probability (%)

0.90
0.80
0.70
0.60

0.50
0.40
0.30
0.20
0.10
0.00
1-14
June

15-30
June

1-14
July

15-31 1-14
July August

15-31
May

2005

1-14
June

15-30
June

1-14
July

15-31 1-14
July August

2006

Fig. 3.5. Relationship between Savannah Sparrow detection probability and ordinal date in CP23 and CP1 Conservation Reserve
Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan, 2005-2006. Error bars indicate 95% confidence
intervals.

136

1.00
0.90

Detection Probability (%)

0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10

0.00
15-31
May

1-14
June

15-30
June

1-14
July

15-31
July

1-14
August

2006

Fig. 3.6. Relationship between Sedge Wren detection probability and ordinal date in
CP23 and CP1 Conservation Reserve Enhancement Program (CREP) grasslands,
Saginaw Bay watershed, Michigan, 2006. Error bars indicate 95% confidence intervals.

137

Table 3.2. Grassland bird species and global occupancy models with corresponding
number of parameters, K, and overdispersion estimate, ĉ, observed in CP23 and CP1
Conservation Reserve Enhancement Program (CREP) grasslands, Saginaw Bay
watershed, Michigan, 2005-2006.
Common Name

Global Model

K

ĉ

Bobolink

Ψ(Field Size), p(Ordinal Date)

4

6.03

Dickcissel

Ψ(CREP Field Type, Field Size), p(.)

4

0.92

Eastern Meadowlark

Ψ(CREP Field Type), p(CREP Field Type)

4

5.49

Grasshopper Sparrow

Ψ(Field Size), p(.)

3

2.78

4

1.66

3

0.67

Sedge Wren, 2006 only Ψ(CREP Field Type), p(Ordinal Date)
Vesper Sparrow

Ψ(CREP Field Type), p(.)

138

Table 3.3. Summary of AIC occupancy model selection adjusted by the best detection
model for Dickcissel in CP23 and CP1 Conservation Reserve Enhancement Program
(CREP) grasslands, Saginaw Bay watershed, Michigan, 2005-2006.
Occupancy Model

-2 log-likelihood

K

AICc

∆AICc

w

Ψ(Field Size), p(.)

98.98

3

107.38

0.00

0.64

104.35

2

109.44

2.06

0.23

98.90

4

111.34

3.96

0.09

104.35

3

112.75

5.37

0.04

Ψ(.), p(.)
Ψ(CREP Field Type, Field Size), p(.)
Ψ(CREP Field Type), p(.)

139

Table 3.4. Parameter estimates for the best fitting occupancy models for Dickcissel and
Grasshopper Sparrow and model averaged parameter estimates for Sedge Wren (2006 data
only) and Vesper Sparrow in CP23 and CP1 Conservation Reserve Enhancement Program
(CREP) grasslands, Saginaw Bay watershed, Michigan, 2005-2006.
95% Confidence
Interval
β
Estimate

Parameter

Dickcissel

Standardized Field Size

Grasshopper Sparrow Intercept

SE

Upper

Lower

-2.21

1.44

-5.02

0.61

0.64

Species

0.58

-0.49

1.77

Sedge Wren

CREP Whole-field Type

-2.25

1.53

-5.25

0.74

Vesper Sparrow

CREP Whole-field Type

-21.83

13.42

-48.13

4.47

140

Table 3.5. Summary of QAIC occupancy model selection adjusted by the best detection
model for Grasshopper Sparrow in CP23 and CP1 Conservation Reserve Enhancement
Program (CREP) grasslands, Saginaw Bay watershed, Michigan, 2005-2006.
-2 log-likelihood

K

QAICc

∆QAICc

w

Ψ(.), p(.)

141.02

2

55.87

0.00

0.84

Ψ(Field Size), p(.)

140.95

3

59.15

3.28

0.16

Occupancy Model

141

Table 3.6. Summary of AIC occupancy model selection adjusted by the best detection
model for Sedge Wren, 2006 data only, in CP23 and CP1 Conservation Reserve
Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan.

Occupancy Model

-2 log-likelihood

K

AICc

∆AICc

w

Ψ(.), p(Ordinal Date)

82.24

3

90.64

0.00

0.50

Ψ(CREP Field Type), p(Ordinal Date)

79.26

4

91.70

1.06

0.30

Ψ(Field Size), p(Ordinal Date)

80.83

4

93.27

2.63

0.13

Ψ(Ordinal Date), p(Ordinal Date)

82.23

4

94.67

4.03

0.07

142

Table 3.7. Summary of AIC occupancy model selection adjusted by the best detection
model for Vesper Sparrow in CP23 and CP1 Conservation Reserve Enhancement
Program (CREP) grasslands, Saginaw Bay watershed, Michigan, 2005-2006.

Occupancy Model

-2 log-likelihood

K

AICc

∆AICc

w

Ψ(.), p(.)

151.34

2

156.43

0.00

0.56

Ψ(CREP Field Type), p(.)

149.24

3

157.64

1.21

0.31

Ψ(Field Size), p(.)

150.95

3

159.35

2.92

0.13

143

APPENDIX 3.2: ALL AVIAN SPECIES OBSERVED

144

Appendix 3.2. Avian species observed in CP23 native fields, the introduced portion of CP1 fields, and CP1 whole fields, Conservation
Reserve Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan, 2005-2006. Avian densities (# individuals /ha
[SE]) are given for species with detection probability estimates ≥0.10. Species observed too rarely to support detection probability
modeling (i.e., observed in < 10% of surveys) or with detection probability estimates <0.10 are noted as present when observed in at least
one survey and as undetected when unobserved.
2005
Common Name

Scientific Name

Native/
CP23

Introduced

2006
CP1

Native/
CP23

Introduced

CP1

Grassland Specialist Species
Bobolink

A

Dolichonyx oryzivorus

0.38 [0.15] 1.33 [0.37]

2.28 [0.93]

0.60 [0.26] 0.68 [0.27]

0.35 [0.15]

Spiza americana

0.04 [0.03] 0.08 [0.07]

0.05 [0.06]

0.03 [0.01] 0.05 [0.03]

0.05 [0.03]

Sturnella magna

0.12 [0.09] 0.004 [0.004] 0.003 [0.003] 0.06 [0.04] 0.01 [0.01]

0.03 [0.02]

Ammodramus savannarum

0.05 [0.02] 0.00 [0.00]

0.00 [0.00]

0.08 [0.03] 0.01 [0.01]

0.01 [0.01]

Ammodramus henslowii

present

undetected

undetected

present

undetected

undetected

Eremophila alpestris

undetected

undetected

undetected

present

undetected

undetected

Circus cyaneus

undetected

present

undetected

present

undetected

undetected

Ring-necked Pheasant

Phasianus colchicus

0.15 [0.10] 0.03 [0.02]

0.07 [0.04]

0.12 [0.07] 0.13 [0.08]

0.07 [0.06]

Savannah Sparrow

Passerculus sandwichensis

0.98 [0.39] 0.37 [0.12]

0.55 [0.17]

2.14 [0.62] 0.94 [0.32]

1.22 [0.46]

Cistothorus platensis

0.01 [0.01] 0.00 [0.00]

present

0.07 [0.02] 0.06 [0.03]

0.03 [0.02]

Pooecetes gramineus

0.07 [0.03] 0.01 [0.01]

0.02 [0.01]

0.06 [0.01] 0.02 [0.01]

0.06 [0.04]

Dickcissel

AB

Eastern Meadowlark
Grasshopper Sparrow
Henslow‘s Sparrow

AB

AC

Horned Lark
Northern Harrier

Sedge Wren

A

Vesper Sparrow

A

145

Appendix 3.2 Continued.
2005
Common Name

Scientific Name

Native/
CP23

Introduced

2006
CP1

Native

Introduced

CP1

Other Species
American Goldfinch

Carduelis tristis

0.33 [0.13] 0.08 [0.03]

0.41 [0.32]

0.24 [0.08] 0.26 [0.07]

0.29 [0.08]

American Robin

Turdus migratorius

undetected

present

undetected

undetected

present

present

American Woodcock

Scolopax minor

undetected

undetected

undetected

present

undetected

undetected

Barn Swallow

Hirundo rustica

NA*

NA*

NA*

present

present

present

Brown Thrasher

Toxostoma rufum

undetected

undetected

undetected

undetected

present

present

Brown-headed Cowbird

Molothrus ater

present

undetected

undetected

present

present

present

Chipping Sparrow

Spizella passerina

undetected

undetected

undetected

undetected

present

present

Clay-colored Sparrow

Spizella pallida

present

present

present

0.05 [0.03] present

present

Common Yellowthroat

Geothlypis trichas

0.03 [0.02] 0.05 [0.04]

0.09 [0.06]

0.12 [0.04] 0.10 [0.06]

0.16 [0.12]

Eastern Kingbird

Tyrannus tyrannus

present

present

present

0.03 [0.01] 0.004 [0.003]

present

House Wren

Troglodytes aedon

present

present

undetected

undetected

undetected

undetected

Indigo Bunting

Passerina cyanea

present

present

present

undetected

present

present

146

Appendix 3.2 Continued.
2005
Native/
CP23

Scientific Name

Killdeer

Charadrius vociferus

present

undetected

undetected

present

undetected

undetected

Mallard

Anas platyrhynchos

present

undetected

undetected

present

undetected

undetected

Cistothorus palustris

undetected

undetected

undetected

present

undetected

undetected

Mourning Dove

Zenaida macroura

present

present

present

present

present

present

Northern Bobwhite Quail

Colinus virginianus

undetected

present

present

undetected

present

undetected

Northern Flicker

Colaptes auratus

present

undetected

undetected

present

undetected

present

Red-winged Blackbird

Agelaius phoeniceus

0.64 [0.26] 2.30 [0.29]

2.18 [0.62]

0.73 [0.19] 2.97 [0.36]

2.11 [0.26]

Song Sparrow

Melospiza melodia

1.54 [0.35] 0.46 [0.08]

1.47 [0.47]

1.89 [0.37] 0.76 [0.14]

1.12 [0.12]

Tree Swallow

Tachycineta bicolor

NA*

NA*

NA*

0.01 [0.01] present

0.06 [0.05]

White-crowned Sparrow

Zonotrichia leucophrys

undetected

undetected

undetected

undetected

undetected

present

Wild Turkey

Meleagris gallopavo

present

undetected

undetected

present

undetected

present

B

A

Michigan Priority Landbird (Knutson et al. 2001, Matteson et al. 2009)
Michigan Species of Special Concern
C
Michigan Endangered Species
* Barn Swallow and Tree Swallow were not included in the 2005 survey.
B

147

CP1

Native/
CP23

Common Name

Marsh Wren

Introduced

2006
Introduced

CP1

APPENDIX 3.3: OCCUPANCY MODELING RESULTS FOR AVIAN SPECIES
OTHER THAN GRASSLAND SPECIALISTS

148

RESULTS
Avian detection and occupancy for non-grassland species were modeled in CP23
and CP1 Conservation Reserve Enhancement Program (CREP) fields. Adequate
detection probability estimates, p, were achieved for 7 species (i.e., p ≥ 0.10), with
detection probability estimates ranging from 0.14 to 0.90 (Table 3.8). For Clay-colored
Sparrow, an adequate detection probability estimate was obtained only when a subset of
the data was used (CP23 2006 data). For Tree Swallow, only data from 2006 could be
used to obtain a detection probability estimate because Tree Swallow data were not
collected in 2005. Avian detection was also modeled in CP23 fields and the introduced
portion of CP1 fields to obtain detection probability estimates (Table 3.8) and enable the
reporting of density information for species in native and introduced vegetation
(Appendix 3.2).
Occupancy modeling was performed for all 7 non-grassland avian species with
adequate detection probability estimates (Table 3.8). The naïve estimate of occupancy
was 1.0 for the detection-only models (i.e., ψ[·] p[covariates]) of American Goldfinch,
Red-winged Blackbird, and Song Sparrow, indicating that these species occupied all
fields in this study. For Common Yellowthroat, the global occupancy model was
extremely overdispersed (i.e., ĉ > 4; Table 3.2), so no further occupancy models were
derived. Global model fit indicated that informative occupancy models could be derived
for the 3 remaining bird species: Clay-colored Sparrow (CP23 2006 data only), Eastern
Kingbird and Tree Swallow (Table 3.9). The best fitting occupancy models for all 3
species did not include occupancy covariates, indicating that none of the covariates tested
in this study were important correlates of the occupancy of Clay-colored Sparrow (CP23

149

fields only), Eastern Kingbird, or Tree Swallow (2005 only) in CP23 and CP1 CREP
fields.
From among three candidate occupancy models, a single best fitting model was
identified for Clay-colored Sparrow (CP23 2006 data only) that included no occupancy
covariates and accounted for 99.8% of the total AICc weight among successful models
(Table 3.10). The naïve estimate of occupancy was 0.57, and the detection corrected
occupancy estimate was 0.58. Beta estimates and 95% confidence intervals for the effect
of the intercept on the best fitting occupancy models of all 3 species for which occupancy
modeling was performed are presented in Table 3.11.
A single best fitting occupancy model was identified from two candidate models
for the Eastern Kingbird data that accounted for 93% of the total AIC c weight and did not
include occupancy covariates (Table 3.12). The naïve estimate of occupancy was 0.64,
and the detection corrected estimates of occupancy was 0.80.
For Tree Swallow in CP23 and CP1 whole CREP fields, a single best fitting
occupancy model that accounted for 69% of the total AICc weight was identified from
among 4 candidate models (Table 3.13). This model did not include occupancy covariates.
The naïve estimate of occupancy was 0.36, and the detection corrected occupancy estimate
was 0.49.

150

Table 3.8. Non-grassland avian species, with corresponding best-fitting detection models
(with constant occupancy, ψ[·]) and detection probability estimates, observed in 1) CP23
and CP1 Conservation Reserve Enhancement Program (CREP) grasslands and in 2) CP23
fields and the introduced portion of CP1 fields, Saginaw Bay watershed, Michigan, 20052006. Ordinal date was standardized to reduce biases in parameter estimation. The
lowest detection probability estimate was reported when multiple detection probability
estimates were generated.
Common Name

Detection Model

p(·) [SE]

CP23 and CP1 Whole CREP Fields
American Goldfinch

p(CREP Field Type)

0.56 [0.06]

Clay-colored Sparrow,
native/CP23 2006 data

p(Ordinal Date)

0.21 [0.14]

Common Yellowthroat

p(Ordinal Date)

0.23 [0.06]

Eastern Kingbird

p(.)

0.14 [0.04]

Red-winged Blackbird

p(CREP Field Type, Ordinal Date)

0.56 [0.09]

Song Sparrow

p(CREP Field Type)

0.90 [0.03]

Tree Swallow, 2005 data

p(.)

0.19 [0.09]

CP23 (Native) and Introduced CP1 Plant Communities
American Goldfinch

p(Grassland Type)

0.42 [0.06]

Clay-colored Sparrow,
native/CP23 2006 data

p(Ordinal Date)

0.21 [0.14]

Common Yellowthroat

p(Ordinal Date)

0.20 [0.05]

Eastern Kingbird, 2006 data

p(.)

0.17 [0.07]

Red-winged Blackbird

p(Grassland Type, Ordinal Date)

0.56 [0.09]

Song Sparrow

p(Grassland Type)

0.84 [0.04]

151

Table 3.9. Non-grassland avian species and global occupancy models with corresponding
number of parameters, K, and overdispersion parameters, ĉ, observed in CP23 and CP1
Conservation Reserve Enhancement Program (CREP) grasslands, Saginaw Bay watershed,
Michigan, 2005-2006.

Common Name

Global Model

K

ĉ

Clay-colored Sparrow
Native 2006 data

Ψ(Field Size), p(Ordinal Date)

4

1.10

Common Yellowthroat

Ψ(Field Size), p(Ordinal Date)

4

4.56

Eastern Kingbird

Ψ(Field Size), p(.)

3

0.41

Tree Swallow

Ψ(CREP Field Type, Field Size), p(.)

4

1.06

152

Table 3.10. Summary of AIC occupancy model selection adjusted by the best detection
model for Clay-colored Sparrow in native/CP23 vegetation of Conservation Reserve
Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan, 2006.

Occupancy Model

-2 log-likelihood

K

AICc

∆AICc

w

Ψ(.), p(Ordinal Date)

39.45

3

53.45

0.00

0.998

Ψ(Native Patch Size), p(Ordinal Date)

38.96

4

66.96

13.50

0.001

Ψ(Ordinal Date), p(Ordinal Date)

39.44

4

67.44

13.99

0.001

153

Table 3.11. Best fitting single occupancy model for Clay-colored Sparrow (Native/CP23
2006 data only), Eastern Kingbird, and Tree Swallow in CP23 and CP1 Conservation
Reserve Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan,
2005-2006
95% Confidence
Interval
Species

Parameter

β
Estimate

SE

Upper

Lower

Clay-colored Sparrow
Native 2006 data

Intercept

0.34

0.79

-1.21

1.89

Eastern Kingbird

Intercept

1.37

1.17

-0.92

3.66

Tree Swallow

Intercept

-0.03

0.97

-1.74

1.68

154

Table 3.12. Summary of AIC occupancy model selection adjusted by the best detection
model for Eastern Kingbird in CP23 and CP1 CREP grasslands, Saginaw Bay watershed,
Michigan, 2006.

-2 log-likelihood

K

AICc

∆AICc

w

Ψ(.), p(.)

106.03

2

113.03

0.00

0.93

Ψ(CREP Field Size), p(.)

104.28

3

118.28

5.25

0.07

Occupancy Model

155

Table 3.13. Summary of AIC occupancy model selection adjusted by the best detection
model for Tree Swallow in CP23 and CP1 Conservation Reserve Enhancement Program
(CREP) grasslands, Saginaw Bay watershed, Michigan, 2005-2006.

Occupancy Model

-2 log-likelihood

K

AICc

∆AICc

w

Ψ(.), p(.)

50.77

2

55.86

0.00

0.69

Ψ(CREP Field Type), p(.)

50.46

3

58.86

3.00

0.15

Ψ(Field Size), p(.)

50.71

3

59.11

3.25

0.14

Ψ(CREP Field Type, Field Size), p(.)

50.45

4

62.89

7.03

0.02

156

CHAPTER 3 LITERATURE CITED

157

Albert, D. A. 1995. Regional landscape ecosystems of Michigan, Minnesota, and
Wisconsin: a working map and classification. General Technical Report NC-178.
St. Paul, Minnesota, USA: U.S. Department of Agriculture, Forest Service, North
Central Forest Experiment Station.

Bakker, K. K., D. E. Naugle, and K. F. Higgins. 2002. Incorporating landscape
attributes into models for migratory grassland bird conservation. Conservation
Biology 16:1638-1646.

Best, L. B., H. Campa, III, K. E. Kemp, R. J. Robel, M. R. Ryan, J. A. Savidge, H. P.,
Weeks, Jr., and S. R. Winterstein. 1997. Bird abundance and nesting in CRP
fields and cropland in the Midwest: a regional approach. Wildlife Society
Bulletin 25:864-877.

Burnham, K. P., and D. R. Anderson. 1998. Model selection and inference: a practical
information-theoretic approach. Springer-Verlag, New York, New York, USA.

Burnham, K. P., and D. R. Anderson. 2002. Model selection and multi-model inference,
2nd Edition. Springer-Verlag, New York, New York, USA.

Cunningham, M. A., and D. H. Johnson. 2006. Proximate and landscape factors
influence grassland bird distributions. Ecological Applications 16:1062-1075.

Edwards, D. K., G. L. Dorsey, and J. A. Crawford. 1981. A comparison of three avian
census methods. Studies in Avian Biology 6:170-176.

Fargione, J. E., T. R. Cooper, D. J. Flaspohler, J. Hill, C. Lehman, T. McCoy, S.
McLeod, E. J. Nelson, K. S. Oberhauser, and D. Tilman. 2009. Bioenergy and
wildlife: threats and opportunities for grassland conservation. BioScience 59:767777.

Fisher, R. J., and S. K. Davis. 2010. From Wiens to Robel: a review of grassland bird
habitat selection. Journal of Wildlife Management 74:265-273.

158

Heard, L. P., A. W. Allen, L. B. Best, S. J. Brady, W. Burger, A. J. Esser, E. Hackett, D.
H. Johnson, R. L. Pederson, R. E. Reynolds, C. Rewa, M. R. Ryan, R. T. Molleur,
and P. Buck. 2000. A comprehensive review of Farm Bill contributions to
wildlife conservation, 1985-2000. W. L. Hohman, and D. J. Halloum, editors.
U.S.D.A., Natural Resources Conservation Service, Wildlife Habitat Management
Institute, Technical Report USDA/NRCS/WHMI-2000.

Herkert, J. R. 2009. Response of bird populations to farmland set-aside programs.
Conservation Biology 23:1036-1040.

Hoekstra, J. M., T. M. Boucher, T. H. Ricketts, and C. Roberts. 2005. Confronting a
biome crisis: global disparities of habitat loss and protection. Ecology Letters
9:23-29.
Hines, J. E. 2006. PRESENCE 2 – Software to estimate patch occupancy and related
parameters. United States Geological Survey – Patuxent Wildlife Research
Center, Laurel, Maryland, USA. http://www.mbr-pwrc.usgs.gov/
software/presence.html. Accessed May 2009.

Johnson, D. H., and L. D. Igl. 2001. Area requirements of grassland birds: a regional
perspective. The Auk 118:24-34.

Knopf, F. L. 1996. Prairie legacies - birds. Pages 115-141 in F. B. Samson and F. L.
Knopf, editors. Prairie conservation: preserving North America‘s most
endangered ecosystem. Island Press, Covelo, California, USA.

Knutson, M. G., G. Butcher, J. Fitzgerald, and J. Shieldcastle. 2001. Partners in Flight
Bird Conservation Plan for The Upper Great Lakes Plain (Physiographic Area
16). USGS Upper Midwest Environmental Sciences Center in cooperation with
Partners in Flight. La Crosse, Wisconsin, USA.

MacKenzie, D. I., J. D. Nichols, G. B. Lachman, S. Droege, J. A. Royle, and C. A.
Langtimm. 2002. Estimating site occupancy rates when detection probabilities
are less than one. Ecology 83:2248–2255.

MacKenzie, D. I., and L. L. Bailey. 2004. Assessing the fit of site-occupancy models.
Journal of Agricultural, Biological, and Environmental Statistics 9:300-318.

159

MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines.
2006. Occupancy estimation and modeling: inferring patterns and dynamics of
species occurrence. Academic Press, Burlington, Massachusetts, USA.

Matteson, S., K. Kreitinger, G. Bartelt, G. Butcher, D. Sample, and T. Will. 2009.
Partners in Flight Bird Conservation Plan for The Boreal Hardwood Transition
(Bird Conservation Region 12 — U.S. Portion). Version 1.0. Partners in Flight.
www.partnersinflight.org. Accessed November 2010.

McCoy, T. D., M. R. Ryan, and L. W. Burger, Jr. 2001. Grassland bird conservation:
CP1 vs. CP2 plantings in Conservation Reserve Program fields in Missouri.
American Midland Naturalist 145:1-17.

McCullagh, P., and J. A. Nelder. 1989. Generalized Linear Models. Chapman and Hall,
New York, New York, USA.

Millenbah, K. F. 1993. The effects of different age classes of fields enrolled in the
Conservation Reserve Program in Michigan on avian diversity, density, and
productivity. M.S. Thesis, Michigan State University, East Lansing, Michigan,
USA.
Nichols, J. D., L. L. Bailey, A. F. O‘Connell, Jr., N. W. Talancy, E. H. Campbell Grant,
A. T. Gilbert, E. M. Annand, T. P. Husband, and J. E. Hines. 2008. Multi-scale
occupancy estimation and modeling using multiple detection methods. Journal of
Applied Ecology 45:1321-1329.

Ribic, C.A., and D. W. Sample. 2001. Associations of grassland birds with landscape
factors in Southern Wisconsin. American Midland Naturalist 146:105-121.

Ribic, C.A., R. R. Koford, J. R. Herkert, D. H. Johnson, N. D. Niemuth, D. E. Naugel, K.
K. Bakker, D. W. Sample, and R. B. Renfrew. 2009. Area sensitivity in North
American grassland birds: patterns and processes. The Auk 126:233-244.

Riffell, S., D. Scognamillo, and L.W. Burger. 2008. Effects of the Conservation Reserve
Program on Northern Bobwhite and grassland birds. Environmental Monitoring
and Assessment 146:309-323.

160

Roloff, G. J., T. E. Grazia, K. M. Millenbah, and A. J. Kroll. In press. Factors associated
with amphibian detection and occupancy in Southern Michigan forests. Journal
of Herpetology.

Sampson, F. B., and F. L. Knopf, editors. 1996. Prairie conservation: preserving North
America‘s most endangered ecosystem. Island Press, Washington, D.C., USA.

Sauer, J. R., B. G. Peterjohn, S. Schwartz, and J. E. Hines. 1995. The grassland bird
home page. Version 95.0. Patuxent Wildlife Research Center, Laurel, Maryland,
USA. http://www.mbr-pwrc.usgs.gov/bbs/grass/grass.htm. Accessed May 2010.

Suter, J. F., G. L. Poe, and N. L. Bills. 2008. Do landowners respond to land retirement
incentives? Evidence from the Conservation Reserve Enhancement Program.
Land Economics 84:17-30.

USDA (U.S. Department of Agriculture). 2000a. Introduced grass planting:
Conservation Reserve Enhancement Program, CREP CP-1. Natural Resources
Conservation Service, Michigan. http://www.michigan.gov/documents/mda_ CP1_INTRODUCEDGRASS_9614_7.pdf. Accessed May 2010.

USDA (U.S. Department of Agriculture). 2000b. Native grass planting: Conservation
Reserve Enhancement Program, CREP CP-2. Natural Resources Conservation
Service, Michigan. http://www.michigan.gov/documents/mda_CP-2_
NATIVEGRASS_9615_7.pdf. Accessed May 2010.

USDA (U.S. Department of Agriculture). 2007. Conservation Reserve Program
summary and enrollment statistics, fiscal year 2006. Farm Service Agency.
http://www.fsa.usda.gov/ Internet/FSA_File/06rpt.pdf. Accessed May 2010.

Wentworth, K. L., M. C. Brittingham, and A. M. Wilson. 2010. Conservation Reserve
Enhancement Program fields: benefits for grassland and shrub-scrub species.
Journal of Soil and Water Conservation 65:50-60.

Winter, M., D. H. Johnson, J. A. Shaffer, T. M. Donovan, and W. D. Svedarsky. 2006.
Patch size and landscape effects on density and nesting success of grassland birds.
Journal of Wildlife Management 70:158-172.

161

CHAPTER 4: GRASSLAND BIRD DETECTION AND OCCUPANCY IN
NATIVE AND INTRODUCED PLANT COMMUNITIES OF MICHIGAN
CONSERVATION RESERVE ENHANCEMENT PROGRAM (CREP)
GRASSLANDS

INTRODUCTION
Grassland bird populations have declined over much of their breeding habitat, and
Breeding Bird Survey (BBS) data indicate that grassland birds have declined more
rapidly, more consistently, and over a wider geographic area than any other guild of
North American birds (Sauer et al. 1995, Knopf 1996). Habitat loss and fragmentation
due to agricultural conversion has been the most important cause of declines (Briggs et al.
2005, Sauer et al. 2008, Fisher and Davis 2010), with losses of native temperate grasslands
>83% in the Midwest. (Noss et al. 1995). Farmland set-aside programs such as the
Conservation Reserve Enhancement Program (CREP) increase the area of native and
non-native grasslands in the U.S., and may diminish or reverse the decline of grassland
bird species (Herkert 2009).
Combinations of within-field and landscape-scale variables have been identified
as important covariates in other studies of grassland birds (e.g., Ribic and Sample 2001,
Bakker et al. 2002, Cunningham and Johnson 2006, Winter et al. 2006), and grassland
birds likely use a hierarchical process that is affected by variables at multiple scales when
choosing breeding territories (Ribic et al. 2009, Fisher and Davis 2010). For example,
Winter et al. (2006) found that grassland bird density was affected by a combination of
habitat patch size, landscape, region, and local vegetation structure more than by local

162

vegetation structure alone. Other research has also demonstrated that the extent and
proximity of woody vegetation influence the distribution and abundance of many
grassland bird species, and larger bird abundances are often associated with less woody
vegetation in the landscape (e.g., Ribic and Sample 2001, Cunningham and Johnson
2006). For example, Cunningham and Johnson (2006) found that tree cover negatively
affected the occurrence of 13 out of 15 observed grassland bird species at proximate
scales (i.e., ≤100 m from birding transects).
Due to geographic variation both in grassland bird habitat requirements (e.g.,
Johnson and Igl 2001, Winter et al. 2006) and in farmland set-aside program
implementation (Heard et al. 2000, Riffell et al. 2008), it is important to evaluate the
impacts of set aside programs on grassland birds at local scales (i.e., state, ecoregional, or
major watershed levels). This study is the first to examine grassland bird occupancy of
Michigan CREP grasslands. The goal of this analysis was to evaluate the correlation of
grassland bird occupancy with 1) structural variables and other characteristics of CREP
grasslands and/or 2) the extent and proximity of nearby (≤100 m) woody vegetation.
Grassland bird occupancy was modeled as a function of these covariates for all observed
grassland bird species that were adequately detected within two different types of CREP
grasslands, native vegetation of CREP conservation practice (CP) 23 plantings and
introduced vegetation of CP1 fields.
METHODS
This study occurred within the 8,219 km2 Sandusky Lake Plain sub-subsection of
Michigan‘s southern Lower Peninsula (Albert 1995). This sub-subsection consists of flat
clay lake plain, bordered by shoreline dunes and sand plain. In the center of the flat clay

163

lake plain, long and narrow till plains and ridges of end moraines parallel either Saginaw
Bay or Lake Huron (Albert 1995). Prior to European settlement, extensive wet and wetmesic prairies occurred upland from the coastal Great Lakes marshes of Saginaw Bay,
but largely due to agricultural conversion, these prairies now occur only as small
remnants, generally on State-owned lands (Albert 1995).
In Michigan, 22,983 ha of land in the Saginaw Bay, Lake Macatawa, and River
Raisin watersheds were enrolled in CREP at the time of this study (USDA 2007). From
CREP lands in the Saginaw Bay watershed, 14 CREP grassland fields in Tuscola County,
ranging in size from 6.8 to 23.6 ha, were selected for study. All grassland study fields
were planted in 2002 and were surveyed for birds in 2005 and 2006. Seven study fields
were CREP conservation practice (CP) type 23 (hereafter CP23) that had been planted
with a seed mixture of native grasses and native and introduced wildflowers over 100%
of the field area, and 7 study fields were CP type 1 (hereafter CP1) fields that had been
planted with introduced grasses and legumes over 70% of the field area, with the
remaining 30% of the field area planted with the same native seed mixture used in CP23
fields (Figure 4.1). In 2005, the native seed mixture cost $16.00/pound (i.e.,
$104.00/acre or $256.99/hectare) and the introduced seed mixture cost $1.90/pound (i.e.,
$20.90/acre or $51.65/hectare) plus 6% sales tax through the Tuscola County
Conservation District. Within CREP, CP23 fields are wetland restorations, and, as
administered in the Saginaw Bay area, all CP23 fields had hydric soils. A small wetland
was created in each CP23 field by cutting drain tiles, and, if necessary, by creating a
push-out, or shallow depression in the soil. For this analysis, I evaluated information
from the upland portion of CP23 fields (hereafter native grasslands) and the introduced

164

portion of CP1 fields (hereafter introduced grasslands). Native grasslands ranged in size
from 6.9 – 19.8 ha (median [SE] = 12.6 [1.9] ha, median = 11.8 ha), and introduced
grasslands ranged in size from 5.5 – 17.0 ha (mean [SE] = 9.4 [1.5], median = 8.0 ha).
Avian data were collected using line-transect sampling (Edwards et al. 1981).
The total upland area of each study grassland was surveyed every 2 weeks from 1 June –
14 August in 2005 (n=5 in all grasslands) and from 15 May – 14 August in 2006 (n=6 per
grassland). Surveys took place along transects running the length of each grassland. The
first transect was established 25 m from a randomly selected grassland corner and
subsequent transects were established at 50 m intervals across the width of the grassland.
An investigator recorded the species and location of each individual bird observed. If a
bird was flushed by the observer, the new location of the bird was noted in order to
prevent counting the same individual bird more than once. If a bird was observed but
could not be identified, the location of the bird was recorded and the species was
designated as unknown. Unidentified individuals were not included in analyses of this
study. Avian surveys were conducted after sunrise and before 12:00 p.m., and surveys
were not performed in the rain or if wind speeds were over 16 kph.
Avian species detection, p, and occupancy, ψ, probabilities in native and
introduced CREP grasslands were estimated using single-season occupancy models
(MacKenzie et al.2002) and the program PRESENCE (Hines 2006). The use of single
season models was based on likelihood ratio testing (MacKenzie et al. 2006, Nichols et
al. 2008), which indicated that the more highly parameterized multi-season models did
not fit the data significantly better than single season models (p > 0.05).

165

Occupancy estimation that is adjusted for imperfect detection is recognized as an
improvement to traditional approaches that assume species are perfectly detected (e.g.,
logistic regression) (Rota et al. 2009). The estimation of a species‘ detection probability
requires multiple surveys at each study site and assumes that the occupancy of each site is
closed to changes between surveys (MacKenzie et al. 2002, MacKenzie et al. 2006).
Concerns about the application of occupancy modeling in ecological systems that violate
this closure assumption have arisen, and violations of closure have been found to
typically result in overestimation of occurrence probability (Rota et al. 2009).
To more closely conform to the closed population assumption, the occupancy data
set for each grassland bird species was truncated based on the estimated date of initial
flight by the first young of the year, or the date at which the population births would
likely be observed during avian surveys. For each species, the date at which initial flight
was likely to occur was estimated from life history information from species accounts in
the Cornell Lab of Ornithology‘s Birds of North America Online (BNA Online) (Lanyon
1995, Martin et al. 1995, Vickery 1996, Giudice and Ratti 2001, Herkert et al. 2001,
Jones and Cornely 2002, Temple 2002, Wheelwright and Rising 2008). Life history
information included arrival time to the breeding grounds, time until nest initiation, time
until incubation, duration of incubation, time until fledging, and, when present, duration
of the post-fledging, pre-flight interval. Estimated dates at which initial flight by the first
young of the year occurred for each grassland species were: Bobolink, June 17;
Dickcissel, July 27; Eastern Meadowlark, June 21; Grasshopper Sparrow, June 17; Ringnecked Pheasant, June 27; Savannah Sparrow, July 1; Sedge Wren, June 18; and Vesper
Sparrow, July 1. For each species, occupancy information collected after the estimated

166

date of initial flight by the first young of the year was not included in detection
probability and occupancy modeling.
For studies with small sample sizes, model ranking using Akaike‘s information
criterion corrected for small sample size (AICc) is recommended (Burnham and
Anderson 2002). The top-ranking detection model for each avian species was identified
as having the lowest AICc value among all candidate models with constant occupancy,
ψ(·) (MacKenzie et al. 2002). Candidate (i.e., successful) detection models were those
that achieved 1) numerical convergence of parameter estimates to >6 significant figures,
2) a balanced variance-covariance matrix, and 3) small- or moderately-sized standard
errors of betas, the untransformed estimates of coefficients for covariates (i.e., standard
error was no greater than 5 times the corresponding beta) (D. MacKenzie, Program
PRESENCE – FAQ at http://www. phidot.org/forum, accessed October 2009).
Avian detection probability was calculated for each species observed in at least
10% of the total number of surveys (MacKenzie et al. 2002). Detection probability
estimates were calculated from the top-ranking detection model for each species.
Potential detection covariates were ordinal date and grassland type (i.e., native or
introduced plant community). Ordinal date was standardized to reduce biases in
parameter estimation (MacKenzie et al. 2002). When grassland type and/or ordinal date
were covariates in the top-ranking detection model, multiple estimates of detection
resulted, and the lowest estimate was reported. For species where the naïve estimate of ψ
equaled 1, study grasslands were fully occupied, and detection modeling was conducted
with ψ fixed to 1.

167

Occupancy modeling was performed only for species with detection probability
estimates ≥0.10. Potential occupancy covariates included grassland type (i.e., native or
introduced), grassland size, and early summer (i.e., 15 May – 15 June) vegetation
variables such as percent bare ground, canopy height, percent total canopy cover, percent
standing dead vegetation, percent grass in the live canopy, percent litter cover, and litter
depth (Table 4.1). Potential occupancy covariates also included two landscape-scale
variables: 1) the total area of woody vegetation within 100 m of study grasslands and 2)
the average proximity of each grassland bird species to woody vegetation (Table 4.1).
Continuous variables were standardized to reduce biases in parameter estimation
(MacKenzie et al. 2002).
Vegetation structure was investigated using one randomly located sampling point
per acre in each CP23 and CP1 field (i.e., 1 sampling point per 0.4 ha; n = 17 to 59
sampling points per field). Vegetation data from CP23 fields and the introduced portion
of CP1 fields were included in occupancy modeling (i.e., data from the native portion of
CP1 fields were omitted). Vegetation data were collected once in 2005 during midsummer (15 June – 14 July) and three times in 2006 during early summer (15 May – 14
June), mid-summer (15 June – 14 July), and late summer (15 July – 14 August). Canopy
height and litter depth were measured at each sampling point using a yardstick. A
modified 50 cm x 25 cm Daubenmire frame was centered on each sampling point
(Daubenmire 1959). All plant species within the Daubenmire frame were identified. The
percentage of Daubenmire frame area composed of bare ground, litter cover, and canopy
cover were visually estimated. Percent standing dead and living vegetation were also
visually estimated, and within the living canopy vegetation, the percent cover of grasses,

168

forbs, and woody vegetation were also determined. The total area of the Daubenmire
frame was accounted for as:

100% plot area = % bare ground + % litter cover + % total canopy cover

where:
% total canopy cover = % dead canopy vegetation + % live canopy vegetation
and:
% live canopy vegetation = % grasses + % forbs + % woody plants.

The percent of woody vegetation within 100 m of each study site was determined
by digitizing land cover images from the National Agricultural Imagery Program (NAIP)
(i.e., 1 m x 1m resolution digital orthophoto quadrangles collected June-August 2005,
http://www.mcgi.state.mi.us/mgdl/DOQQ2005.asp?cnty=Tuscola/) and using ArcGIS
(version 8.2; ESRI 2002) to compute the area of woody vegetation within 100 m buffers
around each grassland. Then the percentage of woody vegetation within each 100m
buffer was calculated as the area of woody vegetation within the buffer divided by the
total area of the buffer. Additionally, the average proximity of each bird species to
woody vegetation was determined for each study using ArcGIS (version 8.2; ESRI 2002),
the digitized woody vegetation data layer, and the location of each individual bird
observed (i.e., the latitude and longitude of each bird). The distance from each observed
bird to the nearest woody vegetation was calculated and these distances were averaged to
obtain the mean proximity of each species to woody vegetation in each study field.
Selection of the set of potential occupancy covariates for each species was based
on variables found to influence abundance or occupancy of the species in other research,
169

including research summarized in the species accounts of BNA Online. Occupancy
covariate selection for each grassland bird species was also informed by results from
CREP administrative-level occupancy modeling (i.e., CP23 and CP1 whole fields;
Chapter 3). Occupancy covariates for Bobolink and Eastern Meadowlark were grassland
type, grassland size, percent litter cover, percent total canopy cover, percent grass in the
live canopy, area of woody vegetation within 100m, and proximity to woody vegetation
(Lanyon 1995, Martin et al. 1995, Ribic and Sample 2001; Table 4.2). For Dickcissel,
occupancy covariates were grassland type, grassland size, percent total canopy cover,
litter depth, area of woody vegetation within 100m, and proximity to woody vegetation
(Temple 2002; Table 4.2).

Occupancy covariates for Grasshopper Sparrow and

Savannah Sparrow were grassland type, grassland size, percent bare ground cover,
percent litter cover, percent total canopy cover, litter depth, area of woody vegetation
within 100m, and proximity to woody vegetation (Vickery 1996, Ribic and Sample 2001,
Bakker et al. 2002, Wheelwright and Rising 2008; Table 4.2). For Ring-necked
Pheasant, occupancy covariates were grassland type, grassland size, and percent standing
dead vegetation in the canopy (Giudice and Ratti 2001; Table 4.2). Occupancy
covariates for Sedge Wren and Vesper Sparrow were grassland type, grassland size,
percent bare ground, percent total canopy cover, canopy height, area of woody vegetation
within 100m, and proximity to woody vegetation (Herkert et al. 2001, Jones and Cornely
2002; Table 4.2)
For each avian species with a minimum detection probability estimate ≥ 0.10, all
possible occupancy models were constructed using the best-fitting detection model as a
function of the occupancy covariates (i.e., ψ[covariates], p[as in best fitting detection

170

model]) (Roloff et al. in press). Candidate occupancy models met the convergence,
variance-covariance matrix, and standard error size criteria as described above. The
global model was identified as the most parameterized occupancy model. Global model
fit (MacKenzie and Bailey 2004, MacKenzie et al. 2006) and top-ranking occupancy
model fit were evaluated using the chi-square goodness-of-fit statistic to estimate an
overdispersion parameter, ĉ, from 2,000 bootstrap iterations. Ecological models are
commonly overdispersed (i.e., ĉ ≠ 1), and adjustment of the model selection criteria to
account for it is advised (MacKenzie and Bailey 2004, MacKenzie et al. 2006). For
species with suitable global models (i.e., ĉ ≤ 4), all occupancy models in the set of
candidate models were ranked according to AICc (Burnham and Anderson 2002). For
species with overdispersed global models, the fit of the top-ranking occupancy model
was evaluated using ĉ, and if the model was not overdispersed, results from the topranking model were reported. When one best occupancy model did not emerge,
competing models (i.e., models where ∆AICc ≤ 2.00) were model averaged to derive a
composite occupancy model (Burnham and Anderson 2002).
RESULTS
Thirty-three avian species were observed in CP23 fields and the introduced
portion of CP1 fields during 2005 and 2006 (Appendix 4.2). In 2005, 23% of individual
birds observed during surveys were unidentified, and in 2006 surveyors were more
experienced and only 8% of observed individuals were unidentified. Using the grassland
bird list developed for the Breeding Bird Survey by the U.S. Geographic Survey (Sauer et
al. 1995), 11 species were identified as grassland specialists: Bobolink (Dolichonyx
oryzivorus), Dickcissel (Spiza americana), Eastern Meadowlark (Sturnella magna),
171

Grasshopper Sparrow (Ammodramus savannarum), Henslow‘s Sparrow (Ammodramus
henslowii), Horned Lark (Eremophila alpestris), Northern Harrier (Circus cyaneus),
Ring-necked Pheasant, Savannah Sparrow (Passerculus sandwichensis), Sedge Wren
(Cistothorus platensis), and Vesper Sparrow (Pooecetes gramineus). Six of these
grassland specialist species are of special conservation priority in Michigan (Appendix
4.2). Bobolink, Dickcissel, Grasshopper Sparrow, Henslow‘s Sparrow, Northern Harrier,
and Sedge Wren are designated as Michigan priority landbirds by the bird conservation
group Partners in Flight (Knutson et al. 2001, Matteson et al. 2009), while Dickcissel,
Grasshopper Sparrow, and Northern Harrier are also state species of special concern and
Henslow‘s Sparrow is a state endangered species (Appendix 4.2).
Bobolink and Savannah Sparrow were the most abundant grassland bird species
in native and introduced grasslands (Appendix 4.2). Three grassland bird species were
too rarely observed to estimate their detection probabilities (i.e., observed in less than
10% of surveys): Henslow‘s Sparrow, Horned Lark, and Northern Harrier. Detection
probability modeling was performed for the 8 other grassland specialist species observed
during this study: Bobolink, Dickcissel, Eastern Meadowlark, Grasshopper Sparrow,
Ring-necked Pheasant, Savannah Sparrow, Sedge Wren, and Vesper Sparrow.
Likelihood ratio testing indicated that more highly parameterized multi-season models
did not fit the data significantly better than simpler single-season models (p > 0.05), so
single season models were used in all detection and occupancy modeling efforts
(MacKenzie et al. 2006, Nichols et al. 2008). In the case of one species, Savannah
Sparrow, the naïve estimate of occupancy equaled 1, so detection modeling was
conducted with ψ fixed equal to 1.

172

Adequate detection probability estimates (i.e., p ≥ 0.10) were obtained for all 8
grassland species, with detection probability estimates ranging from 0.12 to 0.84 (Table
4.3). Ordinal date was included in the best detection model for Dickcissel (Table 4.3).
The detection probability of the species varied between 2005 and 2006, and detection
probability estimates were > 0.12 during both years, indicating that Dickcissel was
adequately detected throughout both breeding seasons (Figure 4.2) The best detection
model for Sedge Wren included grassland type, and the species was adequately detected
in both cover types but tended to be more likely to be detected in native grasslands (p =
0.83 [SE = 0.16] ) than in introduced grasslands (p = 0.14 [SE = 0.16]; Figure 4.3). For
the remaining 6 grassland bird species the best detection model included only the
intercept (Table 4.3).
The naïve estimate of occupancy was 1.0 for Savannah Sparrow, indicating that
all surveyed fields were occupied. Global occupancy model fit was evaluated for the 7
remaining grassland bird species. For Bobolink, global model fit was highly
overdispersed (i.e., ĉ > 4) (Table 4.4), as was the top-ranking occupancy model (ĉ =
8.91), indicating that Bobolink occupancy could not be reliably estimated from this study.
Global model fit indicated that derived occupancy models for the 6 remaining grassland
bird species (i.e., Dickcissel, Eastern Meadowlark, Grasshopper Sparrow, Ring-necked
Pheasant, Sedge Wren (2006 data), and Vesper Sparrow) were not overdispersed and
hence would have informative structure (Table 4.4).
Three competing occupancy models (i.e., ∆AICc ≤ 2) were identified for
Dickcissel that together accounted for 64% of the total AIC c weight among 27 candidate
models (Table 4.5). The top-ranking and second-ranking models were not overdispersed
173

(top-ranked AICc = 87.27, ĉ = 0.87; second-ranked AICc = 87.81, ĉ = 0.86) and included
grassland size (Table 4.5). The second-ranking model also included area of woody
vegetation within 100 m of CREP grasslands (Table 4.5). The third-ranking model was
the intercept-only model (ĉ = 0.87) (Table 4.5). The naïve occupancy estimate was 0.43,
while the detection-corrected occupancy estimates by site ranged from 0.00 to 1.0. The
model averaged beta estimate of the effect of grassland size on Dickcissel occupancy was
negative (-1.94, SE = 1.43; Table 4.6), and the model averaged beta estimate of the effect
of the area of woody vegetation within 100 m of CREP grasslands was positive (beta =
0.18, SE = 0.19; Table 4.6), indicating that Dickcissel occupancy was higher in smaller
grasslands and higher in grasslands with more woody vegetation within 100 m.
However, the 95% confidence interval of both beta estimates included zero, indicating
that grassland type and area of woody vegetation within 100 m were not significant
determinants of Dickcissel occupancy (Table 4.6).
For Eastern Meadowlark, the intercept-only model ranked highest (w = 0.37;
AICc = 29.43; ĉ = 1.52; Table 4.7), indicating that none of the covariates tested in this
study were important correlates of Eastern Meadowlark occupancy. The naïve estimate
of occupancy was 0.14, and the detection corrected occupancy estimate was 0.17.
Two competing (i.e., ∆AICc ≤ 2) occupancy models were identified for
Grasshopper Sparrow that together accounted for 51% of the total AIC c weight among 19
candidate models (Table 4.8). The top-ranking model (AICc = 41.49; ĉ = 0.23) included
area of woody vegetation within 100 m of CREP grasslands, and the second-ranking
model (AICc = 43.11; ĉ = 0.36) was the intercept-only model (Table 4.8). The naïve
174

estimate of occupancy was 0.36, while detection-corrected occupancy estimates by site
ranged from 0.001 to 1.0. The model averaged beta estimate for the effect of the area of
woody vegetation within 100 m of CREP fields was positive (beta = 0.80, SE = 1.07;
Table 4.6) indicating that Grasshopper Sparrow occupancy was higher in grasslands with
more woody vegetation within 100 m. However, the 95% confidence interval of the beta
estimate included zero, indicating that area of woody vegetation within 100 m was not a
significant determinant of Grasshopper Sparrow occupancy (Table 4.6).
For Ring-necked Pheasant, two competing occupancy models (i.e., ∆AICc ≤ 2)
were identified that together accounted for 74% of the total AIC c weight among 5
candidate models (Table 4.9). The intercept-only model (AICc = 67.29; ĉ = 0.25) was the
top-ranking model, and the second-ranking model (AICc = 67.85; ĉ = 0.25) included
grassland type (Table 4.9). The naïve estimate of occupancy was 0.50, and the detectioncorrected occupancy estimate ranged from 0.35 in introduced vegetation to 0.89 in native
vegetation. The model-averaged beta estimate for the effect of grassland type on Ringnecked Pheasant was negative (beta = -2.74, SE = 2.74; Table 4.6), suggesting that native
grasslands had greater occupancy rates than introduced grasslands (i.e., native grasslands
were assigned a value of 0 and introduced vegetation was categorized as 1 during
modeling). The 95% confidence interval of the beta estimate included zero, indicating
that grassland type was not a significant determinant of Ring-necked Pheasant occupancy
(Table 4.6).
Two competing occupancy models (i.e., ∆AICc ≤ 2) were identified for Sedge
Wren that together accounted for 64% of the total AICc weight among 16 candidate
175

models (Table 4.10). The top-ranking model (AICc = 30.75; ĉ = 0.16) did not include
occupancy covariates and the second-ranking model (AICc = 32.13; ĉ = 0.11) included
grassland size (Table 4.10). The naïve occupancy estimate was 0.21, and the detectioncorrected occupancy estimate by site ranged from 0.06 to 0.77. The model averaged beta
estimate for the effect of grassland size on Sedge Wren occupancy was positive (beta =
0.34, SE = 0.35; Table 4.6), indicating that Sedge Wren occupancy was greater in larger
grasslands. However, zero was included in the 95% confidence interval of the beta
estimate, indicating that grassland type was not a significant determinant of Sedge Wren
occupancy (Table 4.6).
For Vesper Sparrow, the intercept-only model was the top-ranking model among
three candidate models (w = 0.63; AICc = 63.06; ĉ = 0.53; Table 4.11), indicating that no
covariates included in this study were important correlates of Vesper Sparrow occupancy.
The naïve estimate of occupancy was 0.50, and the detection-corrected estimate of
occupancy was 0.78.
DISCUSSION
Avian surveys of CP23 and CP1-introduced CREP grasslands in Michigan‘s
Saginaw Bay watershed resulted in the observation of 34 species, 11 of which were
grassland associates (Sauer et al. 1995). Eight grassland bird species were detected
frequently enough to allow detection probability and occupancy modeling. Detection
estimates were adequate (i.e., ≥ 0.10) for all 8 grassland bird species, and were above
0.25 for 5 species, indicating that the survey technique used in this study was an
appropriate methodology for documenting occupancy. Covariates included in the top-

176

ranking occupancy models of grassland bird species included grassland type, grassland
size, and area of woody vegetation within 100 m of study areas. Although the effects of
these covariates were not significant, findings of this study are likely constrained by
small sample sizes, and these covariates warrant evaluation in further research.
Grassland bird species that occurred in this study were similar to those
documented in comparable research. Published studies on grassland bird species in
CREP fields of the Lake States are rare; only one study in Pennsylvania was identified
(Wentworth et al. 2010). Wentworth et al. (2010) found 8 grassland bird species that
were also observed in my study: Bobolink, Dickcissel, Eastern Meadowlark, Grasshopper
Sparrow, Henslow‘s Sparrow, Ring-necked Pheasant, Savannah Sparrow, and Vesper
Sparrow (i.e., Horned Lark, Northern Harrier, and Sedge Wren were not observed in
Pennsylvania but were present in Michigan CREP grasslands). The grassland bird
species that occurred in my study were also similar to those that occurred in comparable
work in grasslands of a similar farmland set-aside program, the Conservation Reserve
Program (CRP). In an investigation of the effects of different age classes of Michigan
CRP fields on avian diversity, density, and productivity, Millenbah (1993) found 10
grassland bird species, all of which were also observed in my study along with Henslow‘s
Sparrow. Nine grassland bird species observed in my surveys also occurred in CRP
fields in a large-scale analysis of Midwestern bird abundance and nesting in CRP fields
and cropland (i.e., Horned Lark and Northern Harrier were absent in CRP fields) (Best et
al. 1997). With the exception of Savannah Sparrow, the grassland species that were
observed in my surveys were also observed in a study of CP1 and switchgrass (Panicum
virgatum) dominated CP2 CRP fields in Missouri (McCoy et al. 2001). These

177

comparisons indicate that grassland bird community composition is similar in CREP and
CRP, however other research suggests that grassland bird abundances vary within and
between CREP and CRP (Chapter 2, McCoy et al. 2001, Wentworth 2010), underscoring
the importance of evaluating CREP and other such conservation programs at local scales
(e.g., within states, ecoregions, or major watersheds). It is important to note that the
consistency of results from studies in different regions may be affected by differences in
grassland composition, age or successional stage, and structure.
The survey methodology of this study generated adequate detection probability
estimates for all grassland species observed in ≥10% of surveys. The technique of
truncating the survey data for each species based upon expected initial flight of the first
young of the year presumably conformed more closely with the closed population
assumption of detection probability and occupancy modeling, and detection probability
estimates for most species were higher for truncated data than for the full set of data (i.e.,
for Bobolink, Eastern Meadowlark, Grasshopper Sparrow, Savannah Sparrow, and Sedge
Wren) (Table 4.3; Chapter 3). In addition to timing surveys to avoid including
population births, other methodological improvements for future avian studies in similar
systems include increasing the number of surveys to improve the probability of detecting
rare species, and increasing the number of fields surveyed to ease or remove analytical
constraints due to small sample sizes.
Despite the constraints of small sample size (i.e., n= 14), the top-ranking or
competing (i.e., ∆AICc ≤ 2) occupancy models of 4 of the 6 grassland bird species with
successful occupancy modeling included covariates. Occupancy modeling results
indicate that grassland type (i.e., native or introduced plant communities) and grassland
178

size may be associated with the occupancy of some grassland bird species in Michigan
CREP grasslands. Grassland type was associated with the occupancy of Ring-necked
Pheasant, with larger occupancy estimates in native grasslands. Grassland size was
associated with both Dickcissel and Sedge Wren occupancy. Higher Sedge Wren
occupancy was associated with larger grasslands, but unexpectedly, higher Dickcissel
occupancy was associated with smaller grasslands, which contrasts with findings of a
study documenting a positive relationship between habitat patch size and Dickcissel
occurrence (Bakker et al. 2002), as well as other research that has shown habitat patch
size to be an important positive influence on the occupancy of Dickcissel and many other
species of grassland birds (summarized in Ribic et al. 2009). Although grassland type
and grassland size were not significant correlates of the occupancy of any species in my
study (i.e., the 95% confidence intervals of the beta estimates for the effect of
field/grassland type and grassland size on occupancy included 0), non-zero tendencies
existed (i.e., 0 occurred near the tails of the 95% confidence intervals of the covariates),
suggesting that these covariates should be included in future work that evaluates
grassland bird occupancy in this or similar ecosystems.
The only other potential occupancy covariate that occurred within top-ranking or
competing occupancy models was area of woody vegetation within 100 m of study
grasslands, which was positively associated with the occupancy of Grasshopper Sparrow
and Dickcissel. However, for both species, zero was present centrally within the 95%
confidence interval of the beta estimate (i.e., within the central 50% of the confidence
interval), indicating that the area of woody vegetation within 100 m of grasslands was not
an important grassland bird occupancy correlate in this study.

179

Other studies have shown that area of woody vegetation around grasslands is an
influence on the occupancy or abundance of grassland birds, often with larger bird
occupancy or abundance estimates associated with less woody vegetation in the
landscape (e.g., Ribic and Sample 2001, Cunningham and Johnson 2006). For example,
in research with similar woody cover digitization methods to those of my study (i.e.,
digitized from 1 m x 1m resolution digital orthophoto quadrangles), Cunningham and
Johnson (2006) demonstrated that tree cover ≤100 m from birding transects negatively
affected the occurrence of Grasshopper Sparrow, Savannah Sparrow, Sedge Wren, and
Bobolink. Results from studies with different methods of quantifying woody vegetation
(e.g., woody vegetation that is mapped at various scales/resolutions) can indicate broadscale trends in the influence of woody vegetation on grassland bird species. In a
summary of over 80 studies pertaining to the effect of woody vegetation on grassland
birds, Bakker (2003) found that investigations that included a metric of the proportion or
increase of woody vegetation in the landscape surrounding grasslands and/or a metric of
the distance to woody vegetation showed negative associations with many grassland
species, including Bobolink, Dickcissel, Eastern Meadowlark, Grasshopper Sparrow,
Ring-necked Pheasant, Savannah Sparrow, and Sedge Wren. Both Bakker (2003) and
Cunningham and Johnson (2006) found a positive association of Vesper Sparrow
occurrence and woody vegetation, indicating that there is variation in the response of
grassland bird species to woody vegetation in the landscape, and suggesting that speciesspecific information should be considered in conservation planning for grassland birds.
Although area of woody vegetation and proximity to woody vegetation were not

180

important covariates of grassland bird occupancy in this study, the results of other
research indicate that these variables should be included in future grassland bird studies.
In a review of 57 studies of grassland bird habitat selection, Fisher and Davis
(2010) identified 9 vegetation features that were important predictors of grassland bird
habitat use (i.e., features correlated with grassland bird occupancy, nest and territory
selection, density, or abundance), which included several vegetation variables that were
potential occupancy covariates in this study: coverage of bare ground, grass, dead
vegetation, and litter, as well as litter depth, and vegetation height. Surprisingly, none of
the top-ranking or competing (i.e., ∆AICc ≤ 2) occupancy models of grassland birds in
my analysis included these variables. In many cases, occupancy models that included
such covariates were unsuccessful (e.g., did not converge), and when such models were
successful, they did not rank as high as null occupancy models or occupancy models that
included other covariates. Other researchers have suggested that in fragmented
landscapes, grassland bird habitat requirements may change in response to habitat
abundance and composition (Andren 1994, Bakker et al. 2002), and it is possible that in
the agriculturally-dominated study system of this project, grasslands were so much more
preferred by grassland birds than rowcrops and other available breeding habitats that
vegetation structural variables were much less important than grassland type and size in
determining grassland bird occupancy.
Management recommendations
Results of this analysis show that CREP grasslands of the Saginaw Bay watershed
provide habitat for a majority of the grassland bird species found in Michigan. Of the 15
grassland species documented in Michigan during development of the most recent
181

Michigan Breeding Bird Atlas (BBAE 2010), 11 species were observed in study areas of
this research project, including all 4 species found to have significantly declining
population trends (p < 0.05) in Michigan in the most recent analysis of grassland bird
Breeding Bird Survey (BBS) data (i.e., Vesper Sparrow, Eastern Meadowlark, Bobolink,
and Ring-necked Pheasant; Sauer et al. 2008). The 4 grassland bird species found in
Michigan during atlas development but not observed in this study were Le Conte‘s
Sparrow, Sharp-tailed Grouse, Short-eared Owl, and Upland Sandpiper.
Although grassland type and grassland size were not significant correlates of the
occupancy of any species, non-zero tendencies existed, suggesting that these covariates
should be included in future studies and considered in grassland bird conservation
strategies. Future studies would benefit from including more study site replicates to
increase sample size and analytical power. Because occupancy covariates differed
among grassland species, I support the recommendations of other researchers that
species-specific habitat preference information should be used when managing grasslands
for grassland bird species (e.g., Murray et al. 2008). For the same reason, I also
recommend that in similar studies the effect of potential occupancy covariates should be
evaluated for individual species within the grassland bird guild rather than for grassland
birds as a group.
Threats to the persistence and increase of farmland set-aside grasslands include
reconversion to corn or other crop. Recently, increased demand for biofuels has resulted
in reconversion of agricultural set-aside lands into crop production (Secchi and Babcock
2007, Searchinger et al. 2008, cited from Fargione et al. 2009), a trend that is likely to
continue due to large, federally mandated increases in biofuel production (i.e., >700%

182

over 2006 production by 2022; Fargione et al. 2009). Fargione et al. (2009) suggest
minimizing the negative effects of increased biofuel demand on wildlife by utilizing
wildlife-compatible biomass sources that do not require use of additional production
lands (e.g., wastes, cover crops, algae), and maximizing the area of perennial grasslands,
including diverse native prairie plantings, to produce biofuel in ways that are compatible
with wildlife. Additionally, I recommend that program administrators provide incentives
for landowners to join or continue in CREP through an increase in the one-time program
sign-up bonus when possible, as this strategy that has been found to generate a greater
increase in CREP enrollment than did smaller increases in the annual payments made
over the term of the CREP contract (Suter et al. 2008). Maintaining and increasing the
area of CREP grasslands should be a grassland bird conservation strategy in Michigan.

183

APPENDIX 4.1: CHAPTER 4 TABLES AND FIGURES

184

CP1 Field
(standard)
n=5

CP23 Field
n=7

Native
Vegetation

Example
Wetland and
25 m Wetland
Buffer

Introduced
Vegetation
(70% area)

CP1 Field
(native planting failed)
n=2
Native
Vegetation
(30% area)

Introduced
Vegetation

Fig. 4.1. Location, type, and number of Conservation Reserve Enhancement Program
(CREP) study fields within Tuscola County, Saginaw Bay watershed, Michigan, 2005
and 2006. Wetlands and wetland buffers were excluded from avian surveys and
vegetation sampling. The surveyed (i.e., upland) area of CP23 fields ranged in size from
6.9 – 19.8 ha (median [SE] = 12.6 [1.9] ha, median = 11.8 ha). The introduced portion of
CP1 fields ranged in size from 5.5 – 17.0 ha (mean [SE] = 9.4 [1.5], median = 8.0 ha),
and CP1 whole fields ranged in size from 7.3 – 23.9 ha (mean [SE] = 12.3 [2.1] ha,
median = 10.9 ha).

185

1.00
0.90

Detection Probability (%)

0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
1-14
June

15-30
June

1-14
July

15-31
July

15-31
May

2005

1-14
June

15-30
June

1-14
July

15-31
July

2006

Fig. 4.2. Relationship between Dickcissel detection probability and ordinal date in CP23 native and CP1 introduced plant
communities of Conservation Reserve Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan, 2005-2006.
Error bars indicate 95% confidence intervals.

186

1.00
0.90

Detection Probability (%)

0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
Native

Introduced

Grassland Plant Community Type
Fig. 4.3. Relationship between Sedge Wren detection probability and grassland type in
CP23 native and CP1 introduced plant communities of Conservation Reserve
Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan, 2005 and
2006. Error bars indicate 95% confidence intervals.

187

Table 4.1. All potential grassland bird occupancy covariates for CP23 native and CP1 introduced plant communities of Conservation
Reserve Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan, 2005-2006.

Early Summer Vegetation Variables

Field

Plant
Community
Type*

Plant
Community
Size (ha)

% Bare
Ground

% Litter
Cover

% Total
Canopy

% Dead
Canopy

% Live Grass Litter
Canopy
Depth (cm)

Canopy
Height (dm)

CP23-1

0

17.78

6.98

68.36

24.66

6.89

6.43

10.20

3.20

CP23-2

0

11.78

9.48

26.03

64.48

1.76

12.40

4.14

3.79

CP23-3

0

6.94

7.06

55.88

37.06

2.90

15.95

7.56

4.29

CP23-4

0

13.32

1.12

76.00

22.88

4.93

9.66

9.35

2.74

CP23-5

0

19.81

12.19

29.79

58.02

5.54

11.39

4.32

3.71

CP23-6

0

7.08

5.00

25.56

69.44

1.98

33.28

4.33

3.78

CP23-7

0

11.47

10.46

42.39

47.14

2.16

16.02

4.95

3.46

CP1-1

1

10.09

3.28

30.32

66.40

0.05

41.33

10.92

7.10

CP1-2

1

6.93

4.69

36.56

58.75

2.64

26.43

5.84

3.50

CP1-3

1

7.60

0.00

18.53

81.47

0.79

30.89

4.76

8.41

CP1-4

1

5.53

0.00

14.71

85.29

0.00

42.42

5.00

6.24

CP1-5

1

7.96

10.68

53.26

36.05

2.36

21.51

4.05

2.63

CP1-6

1

10.90

1.22

70.07

28.70

0.76

19.26

7.20

2.85

CP1-7

1

17.08

2.20

32.45

65.34

0.46

30.45

5.75

6.55

*(0=Native, 1=Introduced)
188

Table 4.1 Continued.
Mean Proximity to Woody Vegetation (m)*

Field

% Woody
Vegetation
within 100m

Bobolink

Dickcissel

Eastern
Meadowlark

Grasshopper
Sparrow

Savannah
Sparrow

Sedge Wren

Vesper
Sparrow

CP23-1

2.34

45.66

.

.

.

82.04

90.16

87.45

CP23-2

4.98

113.57

.

.

118.48

98.69

.

79.77

CP23-3

11.10

.

24.59

.

.

55.87

.

67.49

CP23-4

7.63

.

.

90.50

118.15

134.24

.

.

CP23-5

8.92

303.70

.

.

344.73

202.84

65.37

195.54

CP23-6

9.46

76.67

68.94

89.48

50.80

76.10

.

110.50

CP23-7

10.92

111.64

88.35

.

101.26

108.27

.

.

CP1-1

5.61

181.41

190.79

.

.

153.09

.

80.40

CP1-2

4.78

.

.

.

.

75.23

88.50

.

CP1-3

2.75

118.75

129.82

.

.

86.98

.

.

CP1-4

4.14

158.35

119.20

.

.

137.78

.

.

CP1-5

1.20

132.66

.

.

.

126.68

.

61.84

CP1-6

2.95

148.11

.

.

.

132.11

.

.

CP1-7

10.77

179.53

.

.

.

158.21

.

.

*A period indicates the species was absent in the corresponding field.

189

Table 4.2. Potential occupancy covariates for grassland bird species observed in CP23 native and CP1 introduced plant communities
of Conservation Reserve Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan, 2005-2006.

Eastern
Grasshopper Ring-necked Sedge
Bobolink Dickcissel Meadowlark Sparrow
Pheasant
Wren

Vesper
Sparrow

Plant Community Type

x

x

x

x

x

x

x

Plant Community Size (ha)

x

x

x

x

x

x

x

x

x

x

x

x

x

% Bare Ground

x

% Litter Cover

x

% Total Canopy

x

x
x

x

x

x

% Dead Canopy
% Live Grass Canopy

x
x

Litter Depth (cm)

x
x

x

Canopy Height (dm)
% Woody Vegetation
within 100m

x

x

x

x

x

x

Proximity to Woody
Vegetation (m)

x

x

x

x

x

x

190

Table 4.3. Grassland bird species, with corresponding best-fitting detection models (with
constant occupancy, ψ[·]) and detection probability estimates, observed in CP23 native
and CP1 introduced plant communities of Conservation Reserve Enhancement Program
(CREP) grasslands, Saginaw Bay watershed, Michigan, 2005-2006. The lowest detection
probability estimate was reported when multiple detection probability estimates were
generated.
Common Name

Detection Model

p(·) [SE]

Bobolink

p(.)

0.75 [0.08]

Dickcissel

p(Ordinal Date)

0.12 [0.07]

Eastern Meadowlark

p(.)

0.34 [0.18]

Grasshopper Sparrow

p(.)

0.47 [0.16]

Ring-necked Pheasant

p(.)

0.28 [0.09]

Savannah Sparrow

p(.)

0.84 [0.04]

Sedge Wren

p(Grassland Type)

0.14 [0.16]

Vesper Sparrow

p(.)

0.18 [0.09]

191

Table 4.4. Grassland bird species and global occupancy models with corresponding number of parameters, K, and overdispersion
parameters, ĉ, observed in CP23 native and CP1 introduced plant communities of Conservation Reserve Enhancement Program
(CREP) grasslands, Saginaw Bay watershed, Michigan, 2005-2006.
Common Name

Global Model

K

ĉ

Bobolink

Ψ(Grassland Type, Grassland Size), p(.)

4

8.28

Dickcissel

Ψ(Grassland Type, Grassland Size, Area Woody Vegetation, Litter Depth), p(Ordinal Date)

7

0.88

Eastern Meadowlark

Ψ(Litter Cover, Total Canopy Cover, Live Grass Canopy Cover, Grassland Size), p(.)

6

1.36

Grasshopper Sparrow

Ψ(Bare Ground, Grassland Size, Area Woody Vegetation), p(.)

5

0.33

Ring-necked Pheasant

Ψ(Standing Dead Vegetation, Grassland Size), p(.)

4

0.25

Sedge Wren

Ψ(Total Canopy Cover, Bare Ground, Canopy Height, Grassland Size), p(Grassland Type)

7

0.23

Vesper Sparrow

Ψ(Grassland Size), p(.)

3

0.57

192

Table 4.5. Summary of AIC occupancy model selection adjusted by the best detection model for Dickcissel in CP23 native and CP1
introduced plant communities of Conservation Reserve Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan,
2005-2006.
Occupancy Model

-2 log-likelihood

K

AICc

∆AICc

w

Ψ (Grassland Size, p(Ordinal Date)

74.83

4

87.27

0.00

0.29

Ψ (Grassland Size, Area Woody Vegetation), p(Ordinal Date)

70.31

5

87.81

0.54

0.23

Ψ (.), p(Ordinal Date)

80.72

3

89.12

1.85

0.12

Ψ (Total Canopy Cover), p(Ordinal Date)

77.11

4

89.55

2.28

0.09

Ψ (Total Canopy Cover, Grassland Size), p(Ordinal Date)

73.09

5

90.59

3.32

0.06

Ψ (Grassland Type, Grassland Size), p(Ordinal Date)

73.58

5

91.08

3.81

0.04

Ψ (Litter Depth, Grassland Size), p(Ordinal Date)

74.30

5

91.80

4.53

0.03

Ψ (Area Woody Vegetation), p(Ordinal Date)

79.63

4

92.07

4.80

0.03

Ψ (Total Canopy Cover, Area Woody Vegetation), p(Ordinal Date)

75.56

5

93.06

5.79

0.02

Ψ (Litter Depth), p(Ordinal Date)

80.71

4

93.15

5.88

0.02

Ψ (Grassland Type), p(Ordinal Date)

80.72

4

93.16

5.89

0.02

Ψ (Litter Depth, Grassland Size, Area Woody Vegetation), p(Ordinal Date)

69.73

6

93.73

6.46

0.01

Ψ (Litter Depth, Total Canopy Cover), p(Ordinal Date)

76.25

5

93.75

6.48

0.01

Ψ (Grassland Type, Total Canopy Cover), p(Ordinal Date)

76.45

5

93.95

6.68

0.01

Ψ (Grassland Type, Grassland Size, Area Woody Vegetation), p(Ordinal Date)

70.15

6

94.15

6.88

0.01

Ψ (Grassland Type, Total Canopy Cover, Grassland Size), p(Ordinal Date)

71.18

6

95.18

7.91

0.01

193

Table 4.5 Continued.
Occupancy Model

-2 log-likelihood

K

AICc

∆AICc

w

Ψ (Grassland Type, Litter Depth, Grassland Size), p(Ordinal Date)

71.82

6

95.82

8.55

0.00

Ψ (Litter Depth, Total Canopy Cover, Grassland Size), p(Ordinal Date)

72.11

6

96.11

8.84

0.00

Ψ (Grassland Type, Area Woody Vegetation), p(Ordinal Date)

79.14

5

96.64

9.37

0.00

Ψ (Litter Depth, Area Woody Vegetation), p(Ordinal Date)

79.63

5

97.13

9.86

0.00

Ψ (Grassland Type, Litter Depth), p(Ordinal Date)

80.71

5

98.21

10.94

0.00

Ψ (Grassland Type, Litter Depth, Total Canopy Cover), p(Ordinal Date)

74.37

6

98.37

11.10

0.00

Ψ (Grassland Type, Total Canopy Cover, Area Woody Vegetation),
p(Ordinal Date)

75.54

6

99.54

12.27

0.00

Ψ (Grassland Type, Litter Depth, Total Canopy Cover, Grassland Size),
p(Ordinal Date)

68.23

7

100.90

13.62

0.00

Ψ (Grassland Size, Area Woody Vegetation, Grassland Type, Litter Depth),
p(Ordinal Date)

69.70

7

102.37

15.09

0.00

Ψ (Grassland Type, Litter Depth, Area Woody Vegetation), p(Ordinal Date)

79.14

6

103.14

15.87

0.00

Ψ (Grassland Type, Litter Depth, Total Canopy Cover, Area Woody Vegetation),
p(Ordinal Date)

72.34

7

105.01

17.73

0.00

194

Table 4.6. Best fitting single occupancy models for Dickcissel and Grasshopper Sparrow and composite (i.e., model averaged) occupancy
models for Sedge Wren (2006 data only) and Vesper Sparrow in CP23 native and CP1 introduced plant communities of Conservation
Reserve Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan, 2005-2006.
95% Confidence
Interval
β Estimate

SE

Lower

Upper

Grassland Size
Area Woody Vegetation in 100m Buffer

-1.93
0.18

1.43
0.19

-4.73
-0.19

0.86
0.55

Eastern Meadowlark

Intercept

-1.57

0.84

-3.21

0.08

Grasshopper Sparrow

Area Woody Vegetation in 100m Buffer

0.80

1.07

-1.30

2.90

Ring-necked Pheasant

Grassland Type

-2.74

2.74

-8.11

2.64

Sedge Wren

Grassland Size

0.34

0.35

-0.34

1.03

Vesper Sparrow

Intercept

1.25

1.80

-2.28

4.78

Species

Parameter

Dickcissel

195

Table 4.7. Summary of AIC occupancy model selection adjusted by the best detection model for Eastern Meadowlark in CP23 native
and CP1 introduced plant communities of Conservation Reserve Enhancement Program (CREP) grasslands, Saginaw Bay watershed,
Michigan, 2005-2006.
-2 log-likelihood

K

AICc

∆AICc

w

Ψ(.), p(.)

24.34

2

29.43

0.00

0.37

Ψ(Area Woody Vegetation), p(.)

23.23

3

31.63

2.20

0.12

Ψ(Litter Cover), p(.)

23.80

3

32.20

2.77

0.09

Ψ(Total Canopy Cover), p(.)

24.03

3

32.43

3.00

0.08

Ψ(Grassland Size), p(.)

24.25

3

32.65

3.22

0.07

Ψ(Live Grass Cover), p(.)

24.31

3

32.71

3.28

0.07

Ψ(Litter Cover, Area Woody Vegetation), p(.)

22.41

4

34.85

5.42

0.02

Ψ(Total Canopy Cover, Area Woody Vegetation), p(.)

22.78

4

35.22

5.79

0.02

Ψ(Grassland Size, Area Woody Vegetation), p(.)

22.94

4

35.38

5.95

0.02

Ψ(Litter Cover, Total Canopy Cover), p(.)

23.04

4

35.48

6.05

0.02

Ψ(Live Grass Cover, Area Woody Vegetation), p(.)

23.22

4

35.66

6.23

0.02

Ψ(Litter Cover, Grassland Size), p(.)

23.45

4

35.89

6.46

0.01

Ψ(Litter Cover, Live Grass Cover), p(.)

23.54

4

35.98

6.55

0.01

Ψ(Total Canopy Cover, Grassland Size), p(.)

23.73

4

36.17

6.74

0.01

Ψ(Total Canopy Cover, Live Grass Cover), p(.)

23.85

4

36.29

6.86

0.01

Ψ(Live Grass Cover, Grassland Size), p(.)

24.06

4

36.50

7.07

0.01

Occupancy Model

196

Table 4.7 Continued.
-2 log-likelihood

K

AICc

∆AICc

w

Ψ(Litter Cover, Total Canopy Cover, Area Woody Vegetation), p(.)

20.87

5

38.37

8.94

<0.01

Ψ(Litter Cover, Live Grass Cover. Area Woody Vegetation), p(.)

21.56

5

39.06

9.63

<0.01

Ψ(Total Canopy Cover, Live Grass Cover, Area Woody Vegetation), p(.)

22.15

5

39.65

10.22

<0.01

Ψ(Litter Cover, Total Canopy Cover, Grassland Size), p(.)

22.78

5

40.28

10.85

<0.01

Ψ(Litter Cover, Total Canopy Cover, Live Grass Cover), p(.)

23.04

5

40.54

11.11

<0.01

Ψ(Litter Cover, Live Grass Cover, Grassland Size), p(.)

23.37

5

40.87

11.44

<0.01

Ψ(Total Canopy Cover, Live Grass Cover, Grassland Size), p(.)

23.70

5

41.20

11.77

<0.01

Ψ(Litter Cover, Total Canopy Cover, Live Grass Cover, Grassland Size), p(.)

22.75

6

46.75

17.32

<0.01

Occupancy Model

197

Table 4.8. Summary of AIC occupancy model selection adjusted by the best detection model for Grasshopper Sparrow in CP23 native
and CP1 introduced plant communities of Conservation Reserve Enhancement Program (CREP) grasslands, Saginaw Bay watershed,
Michigan, 2005-2006.
Occupancy Model

-2 log-likelihood K

AICc

∆AICc

w

Ψ(Area Woody Vegetation), p(.)

33.09

3

41.49

0.00

0.35

Ψ(.), p(.)

38.02

2

43.11

1.62

0.16

Ψ(Bare Ground), p(.)

35.26

3

43.66

2.17

0.12

Ψ(Grassland Size), p(.)

36.63

3

45.03

3.54

0.06

Ψ(Bare Ground, Area Woody Vegetation), p(.)

32.71

4

45.15

3.66

0.06

Ψ(Litter Cover, Area Woody Vegetation), p(.)

33.03

4

45.47

3.98

0.05

Ψ(Total Canopy Cover, Area Woody Vegetation), p(.)

33.05

4

45.49

4.00

0.05

Ψ(Total Canopy Cover), p(.)

37.97

3

46.37

4.88

0.03

Ψ(Litter Cover), p(.)

38.00

3

46.40

4.91

0.03

Ψ(Bare Ground, Grassland Size), p(.)

34.85

4

47.29

5.80

0.02

Ψ(Total Canopy Cover, Bare Ground), p(.)

35.25

4

47.69

6.20

0.02

Ψ(Litter Cover, Bare Ground), p(.)

35.25

4

47.69

6.20

0.02

Ψ(Litter Cover, Total Canopy Cover), p(.)

35.25

4

47.69

6.20

0.02

Ψ(Litter Cover, Grassland Size), p(.)

36.43

4

48.87

7.38

0.01

Ψ(Total Canopy Cover, Grassland Size), p(.)

36.60

4

49.04

7.55

0.01

Ψ(Bare Ground, Grassland Size, Area Woody Vegetation), p(.)

32.46

5

49.96

8.47

0.01

198

Table 4.8 Continued.
Occupancy Model

-2 log-likelihood K

AICc

∆AICc

w

Ψ(Litter Cover, Bare Ground, Grassland Size), p(.)

34.78

5

52.28

10.79

<0.01

Ψ(Total Canopy Cover, Bare Ground, Grassland Size), p(.)

34.78

5

52.28

10.79

<0.01

Ψ(Litter Cover, Total Canopy Cover, Grassland Size), p(.)

34.78

5

52.28

10.79

<0.01

199

Table 4.9. Summary of AIC occupancy model selection adjusted by the best detection model for Ring-necked Pheasant in CP23
native and CP1 introduced plant communities of Conservation Reserve Enhancement Program (CREP) grasslands, Saginaw Bay
watershed, Michigan.
Occupancy Model

-2 log-likelihood

K

AICc

∆AICc

w

Ψ(.), p(.)

60.61

2

67.29

0.00

0.42

Ψ(Grassland Type), p(.)

60.61

3

67.85

0.56

0.32

Ψ(Grassland Size), p(.)

60.61

3

69.47

2.18

0.14

Ψ(Standing Dead Vegetation), p(.)

60.61

3

70.22

2.93

0.10

Ψ(Standing Dead Vegetation, Grassland Size), p(.)

60.61

4

73.05

200

5.76

0.02

Table 4.10. Summary of AIC occupancy model selection adjusted by the best detection model for Sedge Wren in CP23 native and
CP1 introduced plant communities of Conservation Reserve Enhancement Program (CREP) grasslands, Saginaw Bay watershed,
Michigan.
Occupancy Model

-2 log-likelihood

K

AICc

∆AICc

w

Ψ(.), p(Grassland Type)

22.35

3

30.75

0.00

0.43

Ψ(Grassland Size), p(Grassland Type)

19.69

4

32.13

1.38

0.21

Ψ(Canopy Height), p(Grassland Type)

21.18

4

33.62

2.87

0.10

Ψ(Bare Ground), p(Grassland Type)

21.24

4

33.68

2.93

0.10

Ψ(Total Canopy Cover), p(Grassland Type)

22.14

4

34.58

3.83

0.06

Ψ(Canopy Height, Grassland Size), p(Grassland Type)

19.05

5

36.55

5.80

0.02

Ψ(Bare Ground, Grassland Size), p(Grassland Type)

19.50

5

37.00

6.25

0.02

Ψ(Total Canopy Cover, Grassland Size), p(Grassland Type)

19.68

5

37.18

6.43

0.02

Ψ(Bare Ground, Canopy Height), p(Grassland Type)

20.37

5

37.87

7.12

0.01

Ψ(Total Canopy Cover, Bare Ground), p(Grassland Type)

20.98

5

38.48

7.73

0.01

Ψ(Total Canopy Cover, Canopy Height), p(Grassland Type)

21.07

5

38.57

7.82

0.01

Ψ(Total Canopy Cover. Canopy Height, Grassland Size),p(Grassland Type)

18.21

6

42.21

11.46

<0.01

Ψ(Bare Ground, Canopy Height, Grassland Size), p(Grassland Type)

18.89

6

42.89

12.14

<0.01

Ψ(Total Canopy Cover, Bare Ground, Grassland Size), p(Grassland Type)

19.49

6

43.49

12.74

<0.01

Ψ(Total Canopy Cover, Bare Ground, Canopy Height), p(Grassland Type)

20.37

6

44.37

13.62

<0.01

Ψ(Total Canopy Cover, Bare Ground, Canopy Height, Grassland Size),
p(Grassland Type)

18.14

7

50.81

20.06

<0.01

201

Table 4.11. Summary of AIC occupancy model selection adjusted by the best detection model for Vesper Sparrow in CP23 native and
CP1 introduced plant communities of Conservation Reserve Enhancement Program (CREP) grasslands, Saginaw Bay watershed,
Michigan.
Occupancy Model

-2 log-likelihood

K

AICc

∆AICc

w

Ψ(.), p(.)

57.97

2

63.06

0.00

0.63

Ψ(Canopy Height), p(.)

56.67

3

65.07

2.01

0.23

Ψ(Grassland Size), p(.)

57.69

3

66.09

3.03

0.14

202

APPENDIX 4.2: ALL AVIAN SPECIES OBSERVED

203

Appendix 4.2. Avian species observed in CP23 native fields, the introduced portion of CP1 fields, and CP1 whole fields, Conservation
Reserve Enhancement Program (CREP) grasslands, Saginaw Bay watershed, Michigan, 2005-2006. Avian densities (# individuals /ha
[SE]) are given for species with detection probability estimates ≥0.10. Species observed too rarely to support detection probability
modeling (i.e., observed in < 10% of surveys) or with detection probability estimates <0.10 are noted as present when observed in at least
one survey and as undetected when unobserved.
2005
Common Name

Scientific Name

Native/
CP23

Introduced

2006
CP1

Native/
CP23

Introduced

CP1

Grassland Specialist Species
Bobolink

A

Dolichonyx oryzivorus

0.38 [0.15] 1.33 [0.37]

2.28 [0.93]

0.60 [0.26] 0.68 [0.27]

0.35 [0.15]

Spiza americana

0.04 [0.03] 0.08 [0.07]

0.05 [0.06]

0.03 [0.01] 0.05 [0.03]

0.05 [0.03]

Sturnella magna

0.12 [0.09] 0.004 [0.004] 0.003 [0.003] 0.06 [0.04] 0.01 [0.01]

0.03 [0.02]

Ammodramus savannarum

0.05 [0.02] 0.00 [0.00]

0.00 [0.00]

0.08 [0.03] 0.01 [0.01]

0.01 [0.01]

Ammodramus henslowii

present

undetected

undetected

present

undetected

undetected

Eremophila alpestris

undetected

undetected

undetected

present

undetected

undetected

Circus cyaneus

undetected

present

undetected

present

undetected

undetected

Ring-necked Pheasant

Phasianus colchicus

0.15 [0.10] 0.03 [0.02]

0.07 [0.04]

0.12 [0.07] 0.13 [0.08]

0.07 [0.06]

Savannah Sparrow

Passerculus sandwichensis

0.98 [0.39] 0.37 [0.12]

0.55 [0.17]

2.14 [0.62] 0.94 [0.32]

1.22 [0.46]

Cistothorus platensis

0.01 [0.01] 0.00 [0.00]

present

0.07 [0.02] 0.06 [0.03]

0.03 [0.02]

Pooecetes gramineus

0.07 [0.03] 0.01 [0.01]

0.02 [0.01]

0.06 [0.01] 0.02 [0.01]

0.06 [0.04]

Dickcissel

AB

Eastern Meadowlark
Grasshopper Sparrow
Henslow‘s Sparrow

AB

AC

Horned Lark
Northern Harrier

Sedge Wren

A

Vesper Sparrow

A

204

Appendix 4.2 Continued.
2005
Common Name

Scientific Name

Native/
CP23

Introduced

2006
CP1

Native

Introduced

CP1

Other Species
American Goldfinch

Carduelis tristis

0.33 [0.13] 0.08 [0.03]

0.41 [0.32]

0.24 [0.08] 0.26 [0.07]

0.29 [0.08]

American Robin

Turdus migratorius

undetected

present

undetected

undetected

present

present

American Woodcock

Scolopax minor

undetected

undetected

undetected

present

undetected

undetected

Barn Swallow

Hirundo rustica

NA*

NA*

NA*

present

present

present

Brown Thrasher

Toxostoma rufum

undetected

undetected

undetected

undetected

present

present

Brown-headed Cowbird

Molothrus ater

present

undetected

undetected

present

present

present

Chipping Sparrow

Spizella passerina

undetected

undetected

undetected

undetected

present

present

Clay-colored Sparrow

Spizella pallida

present

present

present

0.05 [0.03] present

present

Common Yellowthroat

Geothlypis trichas

0.03 [0.02] 0.05 [0.04]

0.09 [0.06]

0.12 [0.04] 0.10 [0.06]

0.16 [0.12]

Eastern Kingbird

Tyrannus tyrannus

present

present

present

0.03 [0.01] 0.004 [0.003]

present

House Wren

Troglodytes aedon

present

present

undetected

undetected

undetected

undetected

Indigo Bunting

Passerina cyanea

present

present

present

undetected

present

present

205

Appendix 4.2 Continued.
2005
Native/
CP23

Scientific Name

Killdeer

Charadrius vociferus

present

undetected

undetected

present

undetected

undetected

Mallard

Anas platyrhynchos

present

undetected

undetected

present

undetected

undetected

Cistothorus palustris

undetected

undetected

undetected

present

undetected

undetected

Mourning Dove

Zenaida macroura

present

present

present

present

present

present

Northern Bobwhite Quail

Colinus virginianus

undetected

present

present

undetected

present

undetected

Northern Flicker

Colaptes auratus

present

undetected

undetected

present

undetected

present

Red-winged Blackbird

Agelaius phoeniceus

0.64 [0.26] 2.30 [0.29]

2.18 [0.62]

0.73 [0.19] 2.97 [0.36]

2.11 [0.26]

Song Sparrow

Melospiza melodia

1.54 [0.35] 0.46 [0.08]

1.47 [0.47]

1.89 [0.37] 0.76 [0.14]

1.12 [0.12]

Tree Swallow

Tachycineta bicolor

NA*

NA*

NA*

0.01 [0.01] present

0.06 [0.05]

White-crowned Sparrow

Zonotrichia leucophrys

undetected

undetected

undetected

undetected

undetected

present

Wild Turkey

Meleagris gallopavo

present

undetected

undetected

present

undetected

present

B

A

Michigan Priority Landbird (Knutson et al. 2001, Matteson et al. 2009)
Michigan Species of Special Concern
C
Michigan Endangered Species
* Barn Swallow and Tree Swallow were not included in the 2005 survey.
B

206

CP1

Native/
CP23

Common Name

Marsh Wren

Introduced

2006
Introduced

CP1

CHAPTER 4 LITERATURE CITED

207

Albert, D. A. 1995. Regional landscape ecosystems of Michigan, Minnesota, and
Wisconsin: a working map and classification. General Technical Report NC-178.
St. Paul, Minnesota, USA: U.S. Department of Agriculture, Forest Service, North
Central Forest Experiment Station.

Andren, H. 1994. Effects of habitat fragmentation of birds and mammals in landscapes
with different proportions of suitable habitat: a review. Oikos 71:355-366.

Bakker, K. K., D. E. Naugle, and K. F. Higgins. 2002. Incorporating landscape
attributes into models for migratory grassland bird conservation. Conservation
Biology 16:1638-1646.

Bakker, K. K. 2003. The effect of woody vegetation on grassland nesting birds: an
annotated bibliography. Proceedings of the South Dakota Academy of Science
82:119-141.

BBAE (Breeding Bird Atlas Explorer online resource). 2010. U.S. Geological Survey
Patuxent Wildlife Research Center & National Biological Information
Infrastructure. http://www.pwrc.usgs.gov/bba. Accessed November 2010.

Best, L. B., H. Campa, III, K. E. Kemp, R. J. Robel, M. R. Ryan, J. A. Savidge, H. P.,
Weeks, Jr., and S. R. Winterstein. 1997. Bird abundance and nesting in CRP
fields and cropland in the Midwest: a regional approach. Wildlife Society
Bulletin 25:864-877.

Briggs, J. M., A. K. Knapp, J. M. Blair, J. L Heisler, G. A. Hoch, M. S. Lett, and J. K.
McCarron. 2005. An ecosystem in transition: causes and consequences of the
conversion of mesic grassland to shrubland. BioScience 55:243-254.

Burnham, K. P., and D. R. Anderson. 2002. Model Selection and Multi-model
Inference, 2nd Edition. Springer-Verlag, New York, New York, USA.

Cunningham, M. A., and D. H. Johnson. 2006. Proximate and landscape factors
influence grassland bird distributions. Ecological Applications 16:1062-1075.

Daubenmire, R. F. 1959. A canopy cover method of vegetational analysis. Northwest
Science 33:43-64.
208

Edwards, D. K., G. L. Dorsey, and J. A. Crawford. 1981. A comparison of three avian
census methods. Studies in Avian Biology 6:170-176.

ESRI. 2002. ArcGIS version 8.2. Earth Research Systems Institute, Redlands,
California, USA.

Fargione, J. E., T. R. Cooper, D. J. Flaspohler, J. Hill, C. Lehman, T. McCoy, S.
McLeod, E. J. Nelson, K. S. Oberhauser, and D. Tilman. 2009. Bioenergy and
wildlife: threats and opportunities for grassland conservation. BioScience 59:767777.

Fisher, R. J., and S. K. Davis. 2010. From Wiens to Robel: a review of grassland bird
habitat selection. Journal of Wildlife Management 74:265-273.

Giudice, J. H., and J. T. Ratti. 2001. Ring-necked Pheasant (Phasianus colchicus), The
Birds of North America Online (A. Poole, Ed.). Cornell Lab of Ornithology,
Ithica, New York, USA. http://bna.birds.cornell.edu/bna/species/176. Accessed
September 2010.

Heard, L. P., A. W. Allen, L. B. Best, S. J. Brady, W. Burger, A. J. Esser, E. Hackett, D.
H. Johnson, R. L. Pederson, R. E. Reynolds, C. Rewa, M. R. Ryan, R. T. Molleur,
and P. Buck. 2000. A comprehensive review of Farm Bill contributions to
wildlife conservation, 1985-2000. W. L. Hohman, and D. J. Halloum, editors.
U.S.D.A., Natural Resources Conservation Service, Wildlife Habitat Management
Institute, Technical Report USDA/NRCS/WHMI-2000.

Herkert, J. R., D. E. Kroodsma, and J. P. Gibbs. 2001. Sedge Wren (Cistothorus
platensis), The Birds of North America Online (A. Poole, Ed.). Cornell Lab of
Ornithology, Ithica, New York, USA. http://bna.birds.cornell.edu/bna/
species/176. Accessed September 2010.

Herkert, J. R. 2009. Response of bird populations to farmland set-aside programs.
Conservation Biology 23:1036-1040.
Hines, J. E. 2006. PRESENCE 2 – Software to estimate patch occupancy and related
parameters. United States Geological Survey – Patuxent Wildlife Research
Center, Laurel, Maryland, USA. http://www.mbrpwrc.usgs.gov/software/presence.html. Accessed May 2009.
209

Johnson, D. H., and L. D. Igl. 2001. Area requirements of grassland birds: a regional
perspective. The Auk 118:24-34.

Jones, S. L., and J. E. Cornely. 2002. Vesper Sparrow (Pooecetes gramineus), The Birds
of North America Online (A. Poole, Ed.). Cornell Lab of Ornithology, Ithica,
New York, USA. http://bna.birds.cornell.edu/bna/species/176. Accessed
September 2010.

Knopf, F. L. 1996. Prairie legacies - birds. Pages 115-141 in F. B. Samson and F. L.
Knopf, editors. Prairie conservation: preserving North America‘s most
endangered ecosystem. Island Press, Covelo, California, USA.

Knutson, M. G., G. Butcher, J. Fitzgerald, and J. Shieldcastle. 2001. Partners in Flight
Bird Conservation Plan for The Upper Great Lakes Plain (Physiographic Area
16). USGS Upper Midwest Environmental Sciences Center in cooperation with
Partners in Flight. La Crosse, Wisconsin, USA.

Lanyon, W. E. 1995. Eastern Meadowlark (Sturnella magna), The Birds of North
America Online (A. Poole, Ed.). Cornell Lab of Ornithology, Ithica, New York,
USA. http://bna.birds.cornell.edu/bna/species/176. Accessed September 2010.

MacKenzie, D. I., J. D. Nichols, G. B. Lachman, S. Droege, J. A. Royle, and C. A.
Langtimm. 2002. Estimating site occupancy rates when detection probabilities
are less than one. Ecology 83:2248–2255.

MacKenzie, D. I., and L. L. Bailey. 2004. Assessing the fit of site-occupancy models.
Journal of Agricultural, Biological, and Environmental Statistics 9:300-318.

MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines.
2006. Occupancy estimation and modeling: inferring patterns and dynamics of
species occurrence. Academic Press, Burlington, Massachusetts, USA.

Martin, S. G., and T. A. Gavin. 1995. Bobolink (Dolichonyx oryzivorus), The Birds of
North America Online (A. Poole, Ed.). Cornell Lab of Ornithology, Ithica, New
York, USA. http://bna.birds.cornell.edu/bna/species/176. Accessed September
2010.

210

Matteson, S., K. Kreitinger, G. Bartelt, G. Butcher, D. Sample, and T. Will. 2009.
Partners in Flight Bird Conservation Plan for The Boreal Hardwood Transition
(Bird Conservation Region 12 — U.S. Portion). Version 1.0. Partners in Flight.
www.partnersinflight.org. Accessed November 2010.

McCoy, T. D., M. R. Ryan, and L. W. Burger, Jr. 2001. Grassland bird conservation:
CP1 vs. CP2 plantings in Conservation Reserve Program fields in Missouri.
American Midland Naturalist 145:1-17.

Millenbah, K. F. 1993. The effects of different age classes of fields enrolled in the
Conservation Reserve Program in Michigan on avian diversity, density, and
productivity. M.S. Thesis, Michigan State University, East Lansing, Michigan,
USA.

Murray, L. D., C. A. Ribic, and W. E. Thogmartin. 2008. Relationship of obligate
grassland birds to landscape structure in Wisconsin. Journal of Wildlife
Management 72:463-467.
Nichols, J. D., L. L. Bailey, A. F. O‘Connell, Jr., N. W. Talancy, E. H. Campbell Grant,
A. T. Gilbert, E. M. Annand, T. P. Husband, and J. E. Hines. 2008. Multi-scale
occupancy estimation and modeling using multiple detection methods. Journal of
Applied Ecology 45:1321-1329.

Noss, R. F., E. T. Laroe, and J. M. Scott. 1995. Endangered ecosystems of the United
States: a preliminary assessment of loss and degradation. National Biological
Service, U.S. Department of the Interior, Report No. 0611-R-01 (MF).
Washington, D. C., USA.

Ribic, C.A., and D. W. Sample. 2001. Associations of grassland birds with landscape
factors in Southern Wisconsin. American Midland Naturalist 146:105-121.

Ribic, C.A., R. R. Koford, J. R. Herkert, D. H. Johnson, N. D. Niemuth, D. E. Naugel, K.
K. Bakker, D. W. Sample, and R. B. Renfrew. 2009. Area sensitivity in North
American grassland birds: patterns and processes. The Auk 126:233-244.

Riffell, S., D. Scognamillo, and L.W. Burger. 2008. Effects of the Conservation Reserve
Program on Northern Bobwhite and grassland birds. Environmental Monitoring
and Assessment 146:309-323.
211

Roloff, G. J., T. E. Grazia, K. M. Millenbah, and A. J. Kroll. In press. Factors associated
with amphibian detection and occupancy in Southern Michigan forests. Journal
of Herpetology.

Rota, C. T., R. J. Fletcher, Jr., R. M. Dorazio, and M. G. Betts. 2009. Occupancy
estimation and the closure assumption. Journal of Applied Ecology 46:11731181.

Sauer, J. R., B. G. Peterjohn, S. Schwartz, and J. E. Hines. 1995. The grassland bird
home page. Version 95.0. Patuxent Wildlife Research Center, Laurel, Maryland,
USA. http://www.mbr-pwrc.usgs.gov/bbs/grass/grass.htm. Accessed May 2010.

Sauer, J. R., J. E. Hines, and J. Fallon. 2008. The North American Breeding Bird
Survey, Results and Analysis 1966 - 2007. Version 5.15.2008. USGS Patuxent
Wildlife Research Center, Laurel, Maryland, USA. http://www.mbrpwrc.usgs.gov/bbs/bbs.html. Accessed November 2010.

Searchinger T., R. Heimlich, R. A. Houghton, F. X. Dong, A. Elobeid, J. Fabiosa, S.
Tokgoz, D. Hayes, and T. H. Yu. 2008. Use of US croplands for biofuels
increases greenhouse gases through emissions from land-use change. Science
319: 1238–1240.

Secchi S. and B. A. Babcock. 2007. Impact of high corn prices on Conservation Reserve
Program acreage. Iowa Ag Review (Spring): 13.

Suter, J. F., G. L. Poe, and N. L. Bills. 2008. Do landowners respond to land retirement
incentives? Evidence from the Conservation Reserve Enhancement Program.
Land Economics 84:17-30.

Temple, S. A. 2002. Dickcissel (Spiza americana), The Birds of North America Online
(A. Poole, Ed.). Cornell Lab of Ornithology, Ithica, New York, USA.
http://bna.birds.cornell.edu/bna/species/176. Accessed September 2010.

USDA (U.S. Department of Agriculture). 2007. Conservation Reserve Program
summary and enrollment statistics, fiscal year 2006. Farm Service Agency.
http://www.fsa.usda.gov/ Internet/FSA_File/06rpt.pdf. Accessed May 2010.

212

Vickery, P. D. 1996. Grasshopper Sparrow (Ammodramus savannarum), The Birds of
North America Online (A. Poole, Ed.). Cornell Lab of Ornithology, Ithica, New
York, USA. http://bna.birds.cornell.edu/bna/species/176. Accessed September
2010.

Wentworth, K. L., M. C. Brittingham, and A. M. Wilson. 2010. Conservation Reserve
Enhancement Program fields: benefits for grassland and shrub-scrub species.
Journal of Soil and Water Conservation 65:50-60.

Wheelwright, N. T., and J. D. Rising. 2008. Savannah Sparrow (Passerculus
sandwichensis), The Birds of North America Online (A. Poole, Ed.). Cornell Lab
of Ornithology, Ithica, New York, USA. http://bna.birds.cornell.edu/bna/species/
176. Accessed September 2010.

Winter, M., D. H. Johnson, J. A. Shaffer, T. M. Donovan, and W. D. Svedarsky. 2006.
Patch size and landscape effects on density and nesting success of grassland birds.
Journal of Wildlife Management 70:158-172.

213

CHAPTER 5: SUMMARY AND MANAGEMENT RECOMMENDATIONS

Results of this study show that CREP grasslands in the Saginaw Bay watershed
provide habitat for grassland bird species in Michigan, including those species of
conservation priority in the State. Grassland birds are one of the most imperiled guilds of
North American birds (Knopf 1996), and populations of grassland bird species continue
to decline in Michigan (Sauer et al. 2008). The most recent population trend analysis of
Breeding Bird Survey (BBS) data from 1980 – 2007 shows significantly decreasing
population trends in Michigan (p < 0.10) for 4 out of 10 grassland bird species included
in the BBS analysis (i.e., Bobolink, Eastern Meadowlark, Ring-necked Pheasant, and
Vesper Sparrow; species included in the analysis were those encountered on ≥ 15 survey
routes), and no species showed significant population increases (Sauer et al. 2008). CP23
and CP1 CREP grasslands of the Saginaw Bay watershed provide habitat for a majority
of grassland birds found in Michigan; of the 15 grassland species documented in
Michigan during development of the most recent Michigan Breeding Bird Atlas (BBAE
2010), 11 species were observed in study areas of this research project, including all 4
species with declining population trends from the BBS analysis (the 4 species found in
Michigan during atlas development but not observed in this study were Le Conte‘s
Sparrow, Sharp-tailed Grouse, Short-eared Owl, and Upland Sandpiper). Six grassland
bird species observed during this study are among those that have been designated as
Michigan priority land birds by the bird conservation group Partners in Flight (Knutson et
al. 2001, Matteson et al. 2009) and have been given management priority by the Michigan
Department of Natural Resources (MDNR). These species are Bobolink, Dickcissel,

214

Grasshopper Sparrow, Henslow‘s Sparrow, Northern Harrier, and Sedge Wren (Knutson
et al. 2001, Matteson et al. 2009), and include 1 Michigan endangered species (i.e.,
Henslow‘s Sparrow) and 3 Michigan special concern species (i.e., Dickcissel, Grasshopper
Sparrow, and Northern Harrier).
Although it was not possible to evaluate the source/sink status of the CREP
grasslands in this study due to unsuccessful efforts to gather avian productivity data, there
is increasing evidence from other studies that farmland set-aside lands are benefiting
grassland bird species at the population level (e.g., Veech 2006, Herkert 2009). Even
small (3-142 ha) grassland fragments have been found to benefit the populations of some
grassland birds, such as Dickcissel and Eastern Meadowlark (Walk et al. 2010),
indicating that the loss of small habitat patches such as the CREP grasslands in this study
could have negative, and perhaps cumulative, impacts on the populations of grassland
birds, and underscoring the importance of maintaining farmland set-aside grasslands such
as those in this study.
There is little doubt that density information is most informative about habitat
quality when coupled with information about reproduction and survival of the species of
interest. Van Horne (1983) raised consciousness that habitat quality as a measure of the
importance of a habitat in maintaining the population of a species ―should be defined in
terms of the survival and production characteristics, as well as the density, of the species
occupying that habitat.‖ However, recently other researchers have provided evidence
that density alone can be a reliable indicator of habitat quality (Bock and Jones 2004,
Perot and Villard 2009). Bock and Jones (2004) surveyed 109 published studies of 67
European or North American bird species and found that habitats with higher densities

215

displayed greater avian reproduction rates (i.e., greater recruitment per capita and per unit
of land area) in the majority of cases (i.e., ≥72%. Bock and Jones (2004) concluded that
―in most cases, density will be a reliable indicator of habitat quality, and bird-count data
will be an appropriate basis from which to make land-management and conservation
decisions.‖
Despite the greater cost of native seed mixtures (i.e., $16.00/pound v.
$1.90/pound for the introduced seed mixture in 2005 in Tuscola County, Michigan) and
the specialized planting techniques required for successful native vegetation
establishment (e.g., use of native seed drills for planting the fluffy seed of some native
grass species), I recommend prioritizing the planting of CP23 whole-field native
grasslands within CREP in the Saginaw Bay area of Michigan. In this study, the
densities of 5 out of 8 grassland bird species that could be included in density
comparisons (i.e., those found in ≥ 10% of surveys and with p ≥ 0.10) were higher in
CP23 fields than in CP1 whole fields and/or the introduced portion of CP1 fields during
at least one year of the study (i.e., Grasshopper Sparrow, Ring-necked Pheasant,
Savannah Sparrow, Sedge Wren, and Vesper Sparrow). The densities of these 5
grassland bird species were higher in CP23 fields than in CP1 whole fields even though
CP1 fields were large enough to support 0 - 22% more grassland bird breeding territories
than CP23 fields, suggesting that CP1 fields containing introduced and native grasslands
are not as attractive to several breeding grassland bird species as similar-sized or smaller
CP23 fields containing only native grasslands. Additionally, the larger amount of
standing dead vegetation in CP23 fields in early summer indicates that the native plant
communities in Michigan CREP grasslands are providing more desirable winter habitat

216

for overwintering wildlife such as Ring-necked Pheasant than are the introduced portions
of CP1 fields (USDA 2000a, 2000b). Although these results suggest that CP23 CREP
grasslands provided preferred habitat for most grassland species, one species of
management priority, Bobolink, was more abundant in the introduced vegetation of CP1
fields during one year of this study. These results demonstrate the diverse response of
grassland birds to habitat features, supporting the recommendations of other researchers
that species-specific habitat preference information should be used when managing
grasslands for grassland bird species (e.g., Murray et al. 2008).
Density comparison results of this study differ from those of other similar studies,
which found no differences between grassland bird abundances in native and introduced
vegetation or greater grassland bird abundances in introduced vegetation (e.g., McCoy et
al. 2001, Wentworth 2010). In these studies, the structure and composition of native
grasslands differed, often widely, from the native fields I examined, tending to be less
diverse, have a denser canopy, and/or to be switchgrass (Panicum virgatum) dominated
(i.e., dense stands of switchgrass have been found to be poor grassland bird habitat;
Norment et al. 1999). In this project, detection probability estimates were adequate (i.e.,
p ≥ 0.10) for all species observed frequently enough to estimate detection (i.e., observed
in ≥ 10% of surveys), indicating that the survey method used in this investigation was
appropriate for detecting grassland bird species in the study system. It is important to
note that the consistency of results from studies in different regions may be affected by
differences in grassland composition, age or successional stage, and structure.
Differences in the findings of this study and of researchers in different regions
underscores the importance of evaluating CREP and other such conservation programs at

217

local scales (e.g., within states, ecoregions, or major watersheds). Local evaluations of
the impacts of farmland set-aside programs on grassland bird populations are also
important because grassland bird habitat requirements vary among geographic regions
(e.g., Johnson and Igl 2001, Winter et al. 2006) and geographic variation in farmland setaside lands exists due to local and regional climatic differences and variation in program
implementation, including management practices (Heard et al. 2000, Riffell et al. 2008).
Occupancy modeling indicated that the occupancy of some grassland bird species
in CREP grasslands of the Saginaw Bay watershed may be associated with CREP field
type (i.e., CP23 or CP1) or grassland type (i.e., native or introduced) as well as
grassland/field size. In CP23 and CP1 CREP whole fields, field type was associated with
the occupancy of Sedge Wren and Vesper Sparrow and field size was associated with
Dickcissel occupancy. In CP23 native and CP1 introduced grasslands, grassland type
was associated with Ring-necked Pheasant occupancy, and grassland size was associated
with Sedge Wren and Dickcissel occupancy. As expected from the coevolutionary
history of native grasses and grassland birds (Sampson and Knopf 1996) and clear
differences in native and introduced plant community composition, higher occupancies of
some grassland bird species (i.e., Sedge Wren, Vesper Sparrow, and Ring-necked
Pheasant) were linked to CP23 fields. Higher Sedge Wren occupancy was associated
with larger grasslands (i.e., CP23 native grasslands ranged from 6.94 – 19.81 ha and CP1
introduced grasslands ranged from 5.53 – 17.01 ha), but unexpectedly, higher Dickcissel
occupancy was associated with smaller fields and grasslands (i.e., see size ranges for
introduced grasslands and CP23 fields or native grasslands above; CP1 whole fields
ranged from 7.32 – 23.9 ha), a finding in contrast with a study documenting a positive

218

relationship between habitat patch size and Dickcissel occurrence (Bakker et al. 2002) as
well as other research that has shown habitat patch size to be an important positive
influence on the occupancy of many other species of grassland birds (summarized in
Ribic et al. 2009). Although field/grassland type and field/grassland size were not
significant correlates of the occupancy of any species in my study (i.e., the 95%
confidence intervals of the beta estimates for the effect of field/grassland type and
field/grassland size on occupancy included 0), non-zero tendencies existed (i.e., 0
occurred near the tails of the 95% confidence intervals of the covariates), suggesting that
these covariates should be included in future work that evaluates grassland bird
occupancy in this or similar ecosystems. Such future studies would benefit from
including more study site replicates to increase sample size and analytical power.
Because occupancy covariates differed among grassland species, I also recommend that
in similar studies the effect of potential occupancy covariates should be evaluated for
individual species within the grassland bird guild rather than for grassland birds as a
group.
Many other potential occupancy covariates were evaluated in addition to
field/grassland type and field/grassland size, but these variables either were not included
in top-ranking/competing models or zero was present centrally within the 95%
confidence interval of the beta estimate (i.e., within the central 50% of the confidence
interval), indicating that they were not important correlates of grassland bird occupancy.
These potential covariates included the average distance to nearest woody vegetation for
each species in each study field, the area of woody vegetation within 100 m of each study
field, and 10 structural variables of grassland vegetation. When choosing breeding

219

territories, grassland birds likely use a hierarchical process that is affected by variables at
multiple scales (Ribic et al. 2009, Fisher and Davis 2010), and combinations of withinfield and landscape-scale variables have been identified as important covariates in other
studies of grassland birds (e.g., Ribic and Sample 2001, Bakker et al. 2002, Cunningham
and Johnson 2006, Winter et al. 2006, Renfrew and Ribic 2008). With one exception,
top-ranking/competing occupancy models did not include more than one covariate, likely
due to constraints caused by small sample size (i.e., n = 14). My findings may have
differed from other research due to analytical restrictions from small sample size (i.e., n =
14) and/or because of differences in the structure or composition of grassland vegetation,
landscape composition, or other study system differences. Based on the results of other
research, information from multiple scales (i.e., within-field and landscape-level
variables) should be included in future grassland bird occupancy or abundance studies
and in conservation management decision making processes.
Other study design improvements are possible in addition to easing or removing
analytical constraints due to small sample size by increasing the number of
fields/grasslands surveyed. The estimation of a species‘ detection probability and
occupancy assumes that the occupancy of each study site is closed to changes between
surveys (MacKenzie et al. 2002, MacKenzie et al. 2006), and violations of the closure
assumption tend to result in overestimates of the probability of occurrence of a species,
which could cause negative consequences when included in conservation and
management decisions for rare or declining species (Rota et al. 2009). To more closely
conform to the closure assumption of occupancy modeling, I recommend using life
history information to truncate the occupancy data set for each grassland species based on

220

the estimated date of initial flight by the first young of the year, or the date by which the
population births would be likely to be observed during avian surveys. Alternatively,
field surveys in breeding habitats could be timed to avoid population births to better
conform to the closed population assumption of detection and occupancy modeling. In
this case, field surveys would need to be conducted only until the latest estimated initial
flight date among species of interest. In my study, detection probability estimates for 5
out of the 8 grassland species that could be included in detection modeling efforts (i.e.,
present in ≥ 10% of surveys) were higher for the truncated data set than for the full set of
data (i.e., 15 May – 15 Aug), providing evidence that this approach is a better
methodological technique. To increase the probability of detection of rare species, I also
recommend maximizing the number of surveys conducted in each study field.
Maintaining and increasing the area of CREP grasslands should be a grassland
bird conservation strategy in Michigan. The results of this study indicate that Michigan
CREP fields, especially CP23 native grasslands, provide grassland bird habitat, a critical
resource for an avian guild whose population declines have occurred more rapidly, more
consistently, and over a wider geographic area than any other guild of North American
birds (Sauer et al. 1995, Knopf 1996), declines that continue to occur in Michigan (Sauer
et al. 2008). When planting new native grasslands, use of Michigan genotypes of native
grass and wildflower species would augment and help maintain local genetic diversity of
grassland plant communities. Also, planting monocultures of switchgrass should be
avoided in grassland bird conservation initiatives, as dense stands of switchgrass have
been found to be poor grassland bird habitat (Norment et al. 1999). Threats to the
maintenance and increase of farmland set-aside grasslands include reconversion of set-

221

aside lands to corn or other crop. Recently, increased demand for biofuels has resulted in
reconversion of agricultural set-aside lands into crop production (Secchi and Babcock
2007, Searchinger et al. 2008, cited from Fargione et al. 2009), a trend that is likely to
continue due to large, federally mandated increases in biofuel production (i.e., >700%
over 2006 production by 2022; Fargione et al. 2009). Fargione et al. (2009) suggest
minimizing the negative effects of increased biofuel demand on wildlife by utilizing
wildlife-compatible biomass sources that do not require use of additional production
lands (e.g., wastes, cover crops, algae), and maximizing the area of perennial grasslands,
including diverse native prairie plantings, to produce biofuel in ways that are compatible
with wildlife (e.g., prohibiting the harvest of biomass during the avian breeding season).
Additionally, I recommend that program administrators provide incentives for
landowners to join or continue in CREP through an increase in the one-time program
sign-up bonus when possible, as this strategy that has been found to generate a greater
increase in CREP enrollment than did smaller increases in the annual payments made
over the term of the CREP contract (Suter et al. 2008). Continued landowner
participation is essential to ensure the lasting presence of grassland bird habitat provided
by farmland set-aside programs such as CREP.

222

CHAPTER 5 LITERATURE CITED

223

Bakker, K. K., D. E. Naugle, and K. F. Higgins. 2002. Incorporating landscape
attributes into models for migratory grassland bird conservation. Conservation
Biology 16:1638-1646.

Bock, C. E., and Z. F. Jones. 2004. Avian habitat evaluation: should counting birds
count? Frontiers in Ecology and the Environment 2:403-410.

BBAE (Breeding Bird Atlas Explorer online resource). 2010. U.S. Geological Survey
Patuxent Wildlife Research Center & National Biological Information
Infrastructure. http://www.pwrc.usgs.gov/bba. Accessed November 2010.

Cunningham, M. A., and D. H. Johnson. 2006. Proximate and landscape factors
influence grassland bird distributions. Ecological Applications 16:1062-1075.

Fargione, J. E., T. R. Cooper, D. J. Flaspohler, J. Hill, C. Lehman, T. McCoy, S.
McLeod, E. J. Nelson, K. S. Oberhauser, and D. Tilman. 2009. Bioenergy and
wildlife: threats and opportunities for grassland conservation. BioScience 59:767777.

Fisher, R. J., and S. K. Davis. 2010. From Wiens to Robel: a review of grassland bird
habitat selection. Journal of Wildlife Management 74:265-273.

Heard, L. P., A. W. Allen, L. B. Best, S. J. Brady, W. Burger, A. J. Esser, E. Hackett, D.
H. Johnson, R. L. Pederson, R. E. Reynolds, C. Rewa, M. R. Ryan, R. T. Molleur,
and P. Buck. 2000. A comprehensive review of Farm Bill contributions to
wildlife conservation, 1985-2000. W. L. Hohman, and D. J. Halloum, editors.
U.S.D.A., Natural Resources Conservation Service, Wildlife Habitat Management
Institute, Technical Report USDA/NRCS/WHMI-2000.

Herkert, J. R. 2009. Response of bird populations to farmland set-aside programs.
Conservation Biology 23:1036-1040.

Johnson, D. H., and L. D. Igl. 2001. Area requirements of grassland birds: a regional
perspective. The Auk 118:24-34.

224

Knopf, F. L. 1996. Prairie legacies - birds. Pages 115-141 in F. B. Samson and F. L.
Knopf, editors. Prairie conservation: preserving North America‘s most
endangered ecosystem. Island Press, Covelo, California, USA.

Knutson, M. G., G. Butcher, J. Fitzgerald, and J. Shieldcastle. 2001. Partners in Flight
Bird Conservation Plan for The Upper Great Lakes Plain (Physiographic Area
16). USGS Upper Midwest Environmental Sciences Center in cooperation with
Partners in Flight. La Crosse, Wisconsin, USA.

MacKenzie, D. I., J. D. Nichols, G. B. Lachman, S. Droege, J. A. Royle, and C. A.
Langtimm. 2002. Estimating site occupancy rates when detection probabilities
are less than one. Ecology 83:2248–2255.

MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines.
2006. Occupancy estimation and modeling: inferring patterns and dynamics of
species occurrence. Academic Press, Burlington, Massachusetts, USA.

Matteson, S., K. Kreitinger, G. Bartelt, G. Butcher, D. Sample, and T. Will. 2009.
Partners in Flight Bird Conservation Plan for the Boreal Hardwood Transition
(Bird Conservation Region 12 — U.S. Portion). Version 1.0. Partners in Flight.
www.partnersinflight.org. Accessed November 2010.

McCoy, T. D., M. R. Ryan, and L. W. Burger, Jr. 2001. Grassland bird conservation:
CP1 vs. CP2 plantings in Conservation Reserve Program fields in Missouri.
American Midland Naturalist 145:1-17.

Murray, L. D., C. A. Ribic, and W. E. Thogmartin. 2008. Relationship of obligate
grassland birds to landscape structure in Wisconsin. Journal of Wildlife
Management 72:463-467.

Norment, C. J., C. D. Ardizzone, and K. Hartman. 1999. Habitat relations and birding
biology of grassland birds in western New York. Studies in Avian Biology
19:112-121.

Perot, A., and M.-A. Villard. 2009. Putting density back into the habitat-quality
equation: case study of an open-nesting forest bird. Conservation Biology
23:1550-1557.

225

Renfrew, R. B., and C. A. Ribic. 2008. Multi-scale models of grassland passerine
abundance in a fragmented system in Wisconsin. Landscape Ecology 23:181193.

Ribic, C.A., and D. W. Sample. 2001. Associations of grassland birds with landscape
factors in Southern Wisconsin. American Midland Naturalist 146:105-121.

Ribic, C.A., R. R. Koford, J. R. Herkert, D. H. Johnson, N. D. Niemuth, D. E. Naugel, K.
K. Bakker, D. W. Sample, and R. B. Renfrew. 2009. Area sensitivity in North
American grassland birds: patterns and processes. The Auk 126:233-244.

Riffell, S., D. Scognamillo, and L.W. Burger. 2008. Effects of the Conservation Reserve
Program on Northern Bobwhite and grassland birds. Environmental Monitoring
and Assessment 146:309-323.

Rota, C. T., R. J. Fletcher, Jr., R. M. Dorazio, and M. G. Betts. 2009. Occupancy
estimation and the closure assumption. Journal of Applied Ecology 46:11731181.

Sampson, F. B., and F. L. Knopf, editors. 1996. Prairie conservation: preserving North
America‘s most endangered ecosystem. Island Press, Washington, D.C., USA.

Sauer, J. R., B. G. Peterjohn, S. Schwartz, and J. E. Hines. 1995. The grassland bird
home page. Version 95.0. Patuxent Wildlife Research Center, Laurel, Maryland,
USA. http://www.mbr-pwrc.usgs.gov/bbs/grass/grass.htm. Accessed May 2010.

Sauer, J. R., J. E. Hines, and J. Fallon. 2008. The North American Breeding Bird
Survey, Results and Analysis 1966 - 2007. Version 5.15.2008. USGS Patuxent
Wildlife Research Center, Laurel, Maryland, USA. http://www.mbrpwrc.usgs.gov/bbs/bbs.html. Accessed November 2010.

Searchinger T., R. Heimlich, R. A. Houghton, F. X. Dong, A. Elobeid, J. Fabiosa, S.
Tokgoz, D. Hayes, and T. H. Yu. 2008. Use of US croplands for biofuels
increases greenhouse gases through emissions from land-use change. Science
319: 1238–1240.

226

Secchi S. and B. A. Babcock. 2007. Impact of high corn prices on Conservation Reserve
Program acreage. Iowa Ag Review (Spring): 13.

Suter, J. F., G. L. Poe, and N. L. Bills. 2008. Do landowners respond to land retirement
incentives? Evidence from the Conservation Reserve Enhancement Program.
Land Economics 84:17-30.

USDA (U.S. Department of Agriculture). 2000a. Introduced grass planting:
Conservation Reserve Enhancement Program, CREP CP-1. Natural Resources
Conservation Service, Michigan. http://www.michigan.gov/documents/mda_ CP1_INTRODUCEDGRASS_9614_7.pdf. Accessed May 2010.

USDA (U.S. Department of Agriculture). 2000b. Native grass planting: Conservation
Reserve Enhancement Program, CREP CP-2. Natural Resources Conservation
Service, Michigan. http://www.michigan.gov/documents/mda_CP-2_
NATIVEGRASS_9615_7.pdf. Accessed May 2010.

Van Horne, B. 1983. Density as a misleading indicator of habitat quality. Journal of
Wildlife Management 47:893-901.

Veech, J.A. 2006. A comparison of landscapes occupied by increasing and decreasing
populations of grassland birds. Conservation Biology 20:1422-1432.

Walk, J. W., E. L. Kershner, T. J. Benson, and R. E. Warner. 2010. Nesting success of
grassland birds in small patches in an agricultural landscape. The Auk 127:328334.

Wentworth, K. L., M. C. Brittingham, and A. M. Wilson. 2010. Conservation Reserve
Enhancement Program fields: benefits for grassland and shrub-scrub species.
Journal of Soil and Water Conservation 65:50-60.

Winter, M., D. H. Johnson, J. A. Shaffer, T. M. Donovan, and W. D. Svedarsky. 2006.
Patch size and landscape effects on density and nesting success of grassland birds.
Journal of Wildlife Management 70:158-172

227