EFFECTS OF OIL DEVELOPMENT ON GRASSLAND SONGBIRDS AND

THEIR AVIAN PREDATORS IN SOUTHEASTERN SASKATCHEWAN

A Thesis

Submitted to the Faculty of Graduate Studies and Research

In Partial Fulfillment of the Requirements

for the Degree of

Master of Science

in

Biology

University of Regina

by

Jason Howard Unruh

Regina, Saskatchewan

October, 2015

Copyright 2015: J.H. Unruh

UNIVERSITY OF REGINA

FACULTY OF GRADUATE STUDIES AND RESEARCH

SUPERVISORY AND EXAMINING COMMITTEE

Jason Howard Unruh, candidate for the degree of Master of Science in Biology, has presented a thesis titled, Effects of Oil Development on Grassland Songbirds and Their Avian Predators in Southeastern Saskatchewan, in an oral examination held on September 29, 2015. The following committee members have found the thesis acceptable in form and content, and that the candidate demonstrated satisfactory knowledge of the subject material. External Examiner: *Dr. Nancy Mahony, Environment Canada

Co-Supervisor: Dr. Mark Brigham, Department of Biology

Co-Supervisor: Dr. Stephen Davis, Adjunct

Committee Member: Dr. Christopher Somers, Department of Biology

Chair of Defense: Dr. Hairuo Qing, Department of Geology *Via Skype

i

ABSTRACT

The quantity and quality of Saskatchewan’s remaining grassland may be

threatened by energy development such as oil extraction. Grassland songbird populations

are declining and increased oil development may be contributing to their declines through

habitat loss and degradation. More quantitative research is needed to inform our

understanding of how grassland songbirds are affected by oil development. I examined

grassland songbird abundance, vegetation structure, habitat type (native and planted

grasslands), and avian predator occurrence across a gradient of oil disturbance to

determine the extent to which oil well proximity, density, and cumulative habitat

disturbance influences the abundance of grassland songbirds and the occurrence of avian

predators. I conducted 486 point counts in 243 sample sites (259 ha) at varying distances

from oil wells, and in areas with varying well densities (0-48 wells/259 ha). The

abundance of seven songbird species was reduced near oil wells or in areas with higher

well densities, the abundance of two species was not influenced by oil wells, and the

abundance of two species increased in the presence of oil wells or with greater well

density. Three species also exhibited reduced abundance with greater cumulative

disturbance, while two species exhibited reduced abundance when the area covered by

well pads or oil access roads increased. I also found evidence that the abundance of four

species was lowest in planted grassland compared to native grassland in the presence of

oil development. My results indicate that oil development influenced vegetation structure,

which likely influenced grassland songbird abundance to some degree. However,

structural changes in vegetation did not account for all observed variation in songbird

abundance. Finally, my results provide evidence that Northern Harrier occurrence is

ii

negatively influenced by oil development but that buteos and corvids are not affected.

Northern Harrier occurrence is possibly influenced by habitat fragmentation caused by oil

development since they are known to be area sensitive. As oil development increases in

grassland habitat, its negative impacts on grassland songbirds will likely become more

pronounced. Efforts should be made to limit well density and the cumulative area of

disturbance on the landscape.

iii

ACKNOWLEDGEMENTS

I thank Dr. Stephen Davis for his guidance and instruction during this entire

process. I am so grateful for the patience, calmness and insight and that Dr. Davis

brought to my research, and for being shaped by his mentoring to become a more

thoughtful, critical and well-rounded biologist. Dr. Davis provided a great deal of

support, time and resources to ensure that I was successful in this endeavor. I also thank

Dr. Mark Brigham for his mentorship, kindness and direction throughout my thesis. Dr.

Brigham successfully coupled a friendly manner with an atmosphere that constantly

challenged and pushed me to become a better thinker and a well-rounded scientist. I fully

enjoyed my time spent in his Bird and Bat Lab and will miss it greatly. I am also

thankful to the past and present members of the Bird and Bat Lab, who provided

distraction and laughter when it was needed, as well as thoughtful insight and discussion.

I am grateful for the cooperation and support from numerous landowners and oil

companies in southeastern SK, who gave permission to conduct research on their

property and allowed access to their rights-of-way. Special thanks to my hard working

and dedicated field crew: D. Sawatzky, S. Ludlow, B. Tether, G. Foley, T. Edkins, L.

Vermeylen, B. Coleman, and R. Guerra.

Finally, this research was made possible by the generous funds and support

provided by a number of sources: The Natural Sciences and Engineering Research

Council of Canada, Saskatchewan Ministry of Environment (Fish and Wildlife

Development Fund), Nature Regina, Nature Saskatchewan, Saskatchewan Wildlife

Federation, Campion College, University of Regina (Faculty of Graduate Studies and

Research), HSE Integrated Ltd.

iv

DEDICATION

I dedicate this thesis to my love and my wife, Laura Unruh. It is hard to express

how grateful I am for you and how integral you were in helping to make this happen.

You have sacrificed so willingly and worked so hard to provide for our family and allow

me to pursue what I am passionate about. You have been so patient, kind and generous

in your encouragement. Thank you for the grace you have extended to me and for

walking with me to end of this journey. I love you.

I also thank my beautiful and radiant daughters: Ava and Emma. Thank you for

loving me as an imperfect Dad, and for welcoming me home every day with a smile.

You challenge me to be a better person, and to forgive with the ease that you forgive me.

Maybe one day you’ll actually read this thesis!

Thank you to my family who have cared for and loved me. Mom and Dad, thank

you for praying for me throughout my life. Thank you for the ways you have shaped me.

Growing up in the Boreal Forest of Ontario with all the camping and outdoor adventuring

we did instilled this passion in me for wildlife and to learn to live in balance with the

natural world. Thank you for also modeling what it means to love and care for people

instead of stuff or money. Janet and Kyle, thank you for your generous support of our

family – we have felt cared for by you.

A special thank you to my friend Brian Wiens. Your encouragement and support

were a constant that I relied on throughout this process. You often brought stability and

perspective to my moments of anxiety. Thank you for your genuine interest in what I do.

You make me feel good about myself and proud of what I have accomplished. You have

a special place in my life.

v

TABLE OF CONTENTS

ABSTRACT.........................................................................................................................i

ACKNOWLEDGMENTS................................................................................................iii

DEDICATION…………………………………………………………………….…….iv

TABLE OF CONTENTS..................................................................................................v

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

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

1. GENERAL INTRODUCTION.....................................................................................1

1.1. Introduction.................................................................................................................1

1.2. Literature Cited..........................................................................................................7

2. EFFECTS OF OIL WELL PROXIMITY, DENSITY AND OVERALL DISTURBANCE ON GRASSLAND SONGBIRD ABUNDANCE.............................12

2.1. Introduction...............................................................................................................12

2.2. Methods......................................................................................................................17

2.2.1. Study area...................................................................................................17

2.2.2. Study site selection.....................................................................................18

2.2.3. Avian and vegetation surveys……………………….….……………….19

2.2.4. Disturbance measurements…………………………….………………..20

2.2.5. Statistical analyses……………………………………….………….…...21

2.2.5.1. Vegetation and oil well models………………….………….…21

2.2.5.2. Bird abundance models………………………….……….……22

2.2.5.3. Landscape disturbance models………………….……….……24

2.3. Results........................................................................................................................25

2.3.1. General Results..........................................................................................25

2.3.2. Vegetation…………………………………………………….………......25

2.3.3. Effects of well proximity, well density and vegetation on bird abundance……………………………………………………………..……...…26

2.3.4. Cumulative disturbance effects on bird abundance…..………..….......29

vi

2.4. Discussion..................................................................................................................30

2.4.1. Effects of well proximity, density and activity on grassland songbird abundance ……………………………………………………………...……….31

2.4.2. Interactive effects of oil wells and grassland type ………………..……33

2.4.3. Cumulative effects of oil development on grassland songbird abundance ………………………………………………………………………34

2.5. Conclusions................................................................................................................38

2.6. Management Implications........................................................................................40

2.7. Literature Cited……………………………………………………………….…………………….42

3. EFFECTS OF OIL WELL DENSITY AND OVERALL DISTURBANCE ON GRASSLAND AVIAN PREDATOR OCCURRENCE..............................................101

3.1. Introduction.............................................................................................................101

3.2. Methods....................................................................................................................103

3.2.1. Study area.................................................................................................103

3.2.2. Study site selection...................................................................................104

3.2.3. Avian surveys…………...........................................................................105

3.2.4. Disturbance measurements………….…………………………………106

3.2.5. Statistical analyses...................................................................................107

3.3. Results......................................................................................................................109

3.3.1. General results.........................................................................................109

3.3.2. Effects of well density and cumulative disturbance on avian predator occurrence...........................................................................................................110

3.4. Discussion................................................................................................................110

3.5. Conclusions..............................................................................................................115

3.6. Management Implications…………………………….………………………….115

3.7. Literature Cited……………………………………….………………………….117

4. GENERAL CONCLUSIONS……………………………..…….............................137

APPENDIX A: SONGBIRD OBSERVATION MODELS…………………………141

vii

APPENDIX B: SONGBIRD VEGETATION MODELS………………….………..154

APPENDIX C: SONGBIRD OIL WELL MODELS …………………….…………165

APPENDIX D: SONGBIRD DISTURBANCE MODELS…………………...……..174

APPENDIX E: AVIAN PREDATOR OBSERVATION MODELS……………….180

viii

LIST OF TABLES

Table 2.1. Distribution of well densities in native and planted sample sites (n=243) in southeastern SK, in 2013-2014. ‘No Well’ = zero wells, ‘Low Density’ = 1-4 wells, ‘Medium Density’ = 5-8 wells, ‘High Density’ = 9+ wells. ‘Active Well’ = total number of sites with an actively producing focal well; ‘Abandoned Well’ = total number of sites with a focal well no longer producing oil.…….…………..52

Table 2.2. Mean species abundance ± SE (with percent occurrence) of eleven commonly

detected grassland songbird species in native and planted pastures in southeastern SK, in 2013 and 2014, and frequency of occurrence for both years and habitat types combined.………………………………………………………...………..53

Table 2.3. Mean values (± SE) for vegetation variables recorded in native and planted

pastures in southeastern SK in 2013-2014, (native=224, planted=262)……...….54 Table 2.4. Vegetation structure models relating distance to oil well (DIST), oil well

density (DENS), and habitat type (native vs. planted; HABITAT) to vegetation parameters in southeastern SK, 2013-2014 (n=486 survey sites). Interactions between habitat type and oil well variables include all main effects. Models for each parameter are shown with the log likelihood score (LL), number of parameters (K), Akaike’s Information Criterion score adjusted for small sample sizes (AICc), ∆AICc, and model weight (wi)……………………………….…….55

Table 2.5. N-mixture abundance models comparing and combining the most

parsimonious oil well models (well proximity [distance], well density [density], well activity [activity]), habitat type [habitat], and vegetation model (native cover [native], exotic grass cover [tame], litter depth [litter], forbs [forbs], shrub distance [shrubs], visual obstruction [robel], bare ground [bare], vegetation height [height]), the observation only model [Detection] and the Null model in southeastern SK, 2013-2014. All oil, vegetation, and habitat models include the observation model. All models are presented with the log likelihood score (LL), number of parameters (K), Akaike’s Information Criterion score adjusted for small sample sizes (AICc), ∆AICc, and model weight (wi)…………………...….58

Table 2.6. Parameter estimates of the top N-mixture model for abundance of eleven

species of grassland songbirds in southeastern SK, 2013-2014……………...….62 Table 2.7. Mean percent of sample site area covered by disturbance features associated

with oil development, divided between well density categories, in southeastern SK, 2013-2014. Total Disturbance is the sum of all other disturbance features…………………………………………………………………………...66

ix

Table 2.8. Correlations between amount of area disturbed in sample sites (n=243) by oil development features in southeastern SK, 2013-2014.…………………………..67

Table 2.9. N-mixture abundance models comparing and combining the most parsimonious landscape level disturbance models (area disturbed by oil roads [oil roads], well pads [well pads], battery and oil building pads [bb pads], pipelines [pipelines], total sum of all disturbance features [total disturbance]), plus habitat type additive and interactive effects, and the observation only model [Detection] and the Null model in southeastern SK, 2013-2014. All disturbance models include the observation model. All models are presented with the log likelihood score (LL), number of parameters (K), Akaike’s Information Criterion score adjusted for small sample sizes (AICc), ∆AICc, and model weight (wi)………....68

Table 2.10. Parameter estimates of the top landscape disturbance N-mixture models of eleven species of grassland songbirds in southeastern SK, 2013-2014.…………71

Table 2.11. Summary table of effects of oil development features and habitat on grassland songbird abundance………………………………...…………………73

Table 3.1. Distribution of well densities in native and tame sample sites (n=243) in southeastern SK, in 2013-2014. ‘No Well’ = zero wells, ‘Low Density’ = 1-4 wells, ‘Medium Density’ = 5-8 wells, ‘High Density’ = 9+ wells. ………...….125

Table 3.2. Mean percent of sample site area taken up by disturbance features associated with oil development, divided between well density categories, in southeastern SK, 2013-2014. Total Disturbance is the sum of all other disturbance features………………………………………………………………………….126

Table 3.3. Correlations between amount of area disturbed in sample sites (n=243) by oil development features in southeastern SK, 2013-2014…………………...……..127

Table 3.4. N-mixture occurrence models comparing and combining the most parsimonious landscape level disturbance models (area disturbed by oil roads [oil roads], well pads [well pads], battery and oil building pads [bb pads], pipelines [pipelines], power line length [power lines], and total sum of all disturbance features [total disturbance]), plus habitat type [habitat], year [year], and the observation only model [Detection] and the Null model in southeastern SK, 2013-2014. All disturbance models include the observation model. All models are presented with the log likelihood score (LL), number of parameters (K), Akaike’s Information Criterion score adjusted for small sample sizes (AICc), ∆AICc, and model weight (wi).…………………………………………………………..…..128

Table 3.5. Parameter estimates of the top landscape disturbance N-mixture occurrence models of three corvid species and three hawk species in southeastern SK, 2013-2014…………………………………………………………………………..…130

x

LIST OF FIGURES Figure 2.1 Map of the study area, grass/pasture area, and 243 sample sites (with well

densities) in southeastern SK, 2013-2014……………………………….…….....75 Figure 2.2 Typical sample site showing the 908 m, 400 m and 100 m buffers, along with

the locations of avian point counts and oil wells…………….…………………..76

Figure 2.3. Predicted change in proportion of bare ground cover (± 85% CI) in relationship to well proximity in southeastern SK, 2013-2014………………….77

Figure 2.4. Predicted change in proportion of native cover (± 85% CI) in relationship to well proximity in southeastern SK, 2013-2014………………………………….78

Figure 2.5. Predicted change in proportion of exotic grass cover (± 85% CI) in relationship to well proximity in southeastern SK, 2013-2014………………….79

Figure 2.6. Predicted change in visual obstruction and density (Robel; ± 85% CI) in relationship to well density and well proximity in southeastern SK, 2013-2014……………………………………………………………………………....80

Figure 2.7. Predicted change in vegetation height (± 85% CI) in relationship to well density and well proximity in southeastern SK, 2013-2014…………….……….81

Figure 2.8. Predicted change in distance to the nearest shrub (± 85% CI) in relationship to oil well proximity in southeastern SK, 2013-2014………………………...…….82

Figure 2.9. Model-predicted mean abundance (± 85% CI) of Baird’s Sparrow varied with well proximity at active and abandoned well sites in native and planted pastures in southeastern SK, 2013-2014.…………………………………………………….83

Figure 2.10. Model-predicted mean abundance (± 85% CI) of Baird’s Sparrow varied with well activity between native and planted pastures, and with litter depth, forb cover, and exotic grass cover in southeastern SK, 2013-2014.……………………………………………………………………….……..84

Figure 2.11. Model-predicted mean abundance (± 85% CI) of Brown-headed Cowbird varied with well activity between native and planted pastures, and with visual obstruction in southeastern SK, 2013-2014.……………………………………..85

Figure 2.12. Model-predicted mean abundance (± 85% CI) of Bobolink varied with well distance between native and planted pastures, and with vegetation height, shrub distance, and exotic grass cover in southeastern SK, 2013-2014…….………….86

Figure 2.13. Model-predicted mean abundance (± 85% CI) of Chestnut-collared Longspurs varied with well proximity, habitat type, litter depth, forb cover, and exotic grass cover in southeastern SK, 2013-2014………………………..……..87

Figure 2.14. Model-predicted mean abundance (± 85% CI) of Grasshopper Sparrows in 2013 varied with well proximity and visual obstruction in southeastern SK……88

xi

Figure 2.15. Model-predicted mean abundance (± 85% CI) of Grasshopper Sparrows in 2014 varied with well proximity, well activity, and litter depth and shrub distance in southeastern SK…………...…………………………………………………..89

Figure 2.16. Model-predicted mean abundance (± 85% CI) of Savannah Sparrow varied with well proximity, visual obstruction, shrub distance, and bare ground cover in southeastern SK, 2013-2014……….…………………………………...………..90

Figure 2.17. Model-predicted mean abundance (± 85% CI) of Vesper Sparrow varied with well proximity, visual obstruction, and native and exotic grass cover in southeastern SK, 2013-2014.…………………………………………………….91

Figure 2.18. Model-predicted mean abundance (± 85% CI) of Sprague’s Pipit varied with well density, habitat type, litter depth, native cover, and visual obstruction in southeastern SK, 2013-2014.…………………………………………………….92

Figure 2.19. Model-predicted mean abundance (± 85% CI) of Western Meadowlark varied with well density in native and planted pastures, litter depth, and visual obstruction in southeastern SK, 2013-2014.……………………………………..93

Figure 2.20. Model-predicted mean abundance (± 85% CI) of Clay-colored Sparrow

varied with shrub distance, visual obstruction and litter depth in southeastern SK, 2013-2014………………..………………………………………………………94

Figure 2.21. Model-predicted mean abundance (± 85% CI) of Horned Lark varied with percent litter depth, visual obstruction, and habitat type in southeastern SK, 2013-2014.……………………………………………………………………..……….95

Figure 2.22. Correlation between well density and total area disturbed by oil development in samples (n=243) in southeastern SK, 2013-2014……………………………………………………………………...……….96

Figure 2.23. Model-predicted mean abundance (± 85% CI) of Bobolink, Clay-colored Sparrow, Grasshopper Sparrow and Savannah Sparrow at the landscape scale varied with total disturbance in southeastern SK, 2013-2014.…………..………97

Figure 2.24. Model-predicted mean abundance (± 85% CI) of Baird’s Sparrow at the landscape scale varied with percent area disturbed by well pads in native and planted pastures in southeastern SK, 2013-2014.…………………………..……98

Figure 2.25. Model-predicted mean abundance (± 85% CI) of Grasshopper Sparrow at the landscape scale varied with percent area disturbed by battery and oil building pads in southeastern SK, 2013-2014.………………………………………………….99

Figure 2.26. Model-predicted mean abundance (± 85% CI) of Western Meadowlark at the landscape scale varied with percent area disturbed by oil roads in native and planted pastures in southeastern SK, 2013-2014.………………….………….100

xii

Figure 3.1. Map of the study area, grass/pasture area, and 243 sample sites (with well densities) in southeastern SK, 2013-2014………………………...…………….132

Figure 3.2. Typical sample site showing the 908 m and 100 m radius buffers, along with the locations of the avian point count and oil wells………………...…………..133

Figure 3.3. Correlation between well density and total disturbed area in sample sites (n=243) in southeast SK, 2013-2014……………………………….…………..134

Figure 3.4. Model predicted mean occurrence (± 85% CI) of Northern Harriers varied by well density in southeastern SK, 2013-2014……………………………………135

Figure 3.5. Model predicted mean occurrence (± 85% CI) of American Crows varied by habitat type and year in southeastern SK, 2013-2014……………………..……136

1

1. GENERAL INTRODUCTION

1.1 INTRODUCTION

There is an ongoing conflict between resource development to meet the needs of

our preferred lifestyle and our desire for clean air and water, and protection of the flora

and fauna around us. Unfortunately, the more we exploit any resource the more habitat

is likely to be altered or destroyed and wildlife populations are reduced or become extinct.

There is an urgent need to understand the full extent of possible habitat destruction and

wildlife population reduction caused by resource development and to strike a balance

between habitat and wildlife conservation and further resource development.

One of the most human-altered landscapes on the planet is the Great Plains of

North America. These grasslands are considered the most altered and threatened

ecosystem in North America (Hickman et al. 2006, Askins et al. 2007, Fisher and Davis

2011). Canada has an estimated 20% of its original native grasslands remaining (Samson

and Knopf 1994). In Manitoba, 99.9% of tallgrass and 61% of mixed grass prairie has

been lost (Samson and Knopf 1994), the same amount of mixed grass prairie has been

lost in Alberta (Samson and Knopf 1994), while in Saskatchewan only 21%

(approximately 5 million ha.) of native prairie remains (Hammermeister et al. 2001).

Most grassland loss can be attributed to conversion to cropland; however, resource

extraction has also contributed to grassland habitat loss and degradation in recent decades.

This is of concern because native grasslands in Saskatchewan provide breeding habitat

for many avian species of conservation concern (Henderson and Davis 2014).

Grassland birds in North America are of high conservation concern. Grassland

birds have exhibited steeper declines than any other group of birds in North America in

recent decades, with average population declines of 40% since 1970 (North American

2

Bird Conservation Initiative Canada 2012). Some species, such as Sprague’s Pipit

(Anthus spragueii), Baird’s Sparrow (Ammodramus bairdii), Chestnut-collared Longspur

(Calcarius ornatus), and McCown’s Longspur (Rhynchophanes mccownii) have

exhibited declines of 68-91% since 1970 (North American Bird Conservation Initiative

U.S. Committee 2009). Overall, 22 of 28 obligate grassland species for which there are

sufficient data, are in decline (Sauer et al. 2014). As a result, declines of North American

grassland birds have emerged as a prominent conservation issue of the 21st century

(Brennan and Kuvlesky 2005).

The major causes typically cited for grassland bird declines are conversion of

native grassland to cropland, afforestation, fragmentation, human settlement, fire

suppression, invasion by exotic grasses, and large-scale deterioration of rangelands

(Murphy and Moore 2003, Brennan and Kuvlesky 2005, Askins et al. 2007). However,

infrastructure associated with the extraction of oil (henceforth oil development) may also

be threatening the quality and quantity of the remaining grassland habitat in

Saskatchewan.

Oil extraction is a common activity in North American grasslands and is rapidly

expanding. Saskatchewan had an estimated 30,756 productive oil wells in 2014

(Government of Saskatchewan: Economy 2015), with thousands of new wells being

drilled every year. Even though oil extraction on the prairies has been ongoing for

decades, few studies have examined the effects of oil development on grassland

songbirds, and the cumulative effects of industrial development on endemic bird

communities is poorly understood (Linnen 2008, Dale et al. 2009, Souther et al. 2014).

Research that has assessed the effects of other types of energy development, such as wind

3

energy, have found reduced grassland bird abundance and increased mortality associated

with these development features (Leddy et al. 1999, Smallwood 2013, Stevens et al.

2013). Furthermore, studies conducted in the boreal forest found reduced songbird

pairing success (Habib et al. 2007) and abundance (Bayne et al. 2005, Bayne et al. 2008,

Bayne and Dale 2011) associated with oil development. Thus, it is reasonable to expect

that oil development negatively impacts grassland songbirds.

The presence of oil wells and associated disturbances may negatively affect the

abundance of grassland songbirds. Birds may avoid oil wells and associated disturbances

by not incorporating them into their territories. Results from the few studies that have

evaluated grassland songbird response to oil development suggest that species such as

Baird’s Sparrow, Chestnut-collared Longspurs, Sprague’s Pipit, and Western

Meadowlark (Sturnella neglecta) avoid areas near oil wells and oil disturbance features

(Linnen 2008, Lawson et al. 2011, Ludlow et al. 2015). This response may drive

landscape-level effects, whereby grassland songbird abundance may be lower in areas

with high well densities than areas with few or no oil wells. Gilbert and Chalfoun (2011)

found that Vesper Sparrow (Pooecetes gramineus), Brewer’s Sparrow (Spizella breweri)

and Sage Sparrow (Artemisiospiza nevadensis) had reduced abundance associated with

increased well density in sagebrush-steppe habitat. Furthermore, areas with higher

cumulative disturbance associated with oil extraction would be expected to have lower

bird abundance than areas with fewer disturbances. The amount of cumulative

disturbance (e.g. roads, pipelines, exotic vegetation, buildings and batteries) associated

with oil extraction may negatively influence abundance or occurrence of grassland

songbirds. Increased disturbance, particularly linear features such as roads and pipelines,

4

may degrade habitat and subsequently reduce bird abundance (Davis 2004, Ribic et al.

2009). While the exact mechanisms are unknown, avoidance of oil wells and associated

disturbance features may result from changes in vegetation structure surrounding these

sites (Linnen 2008, Kalyn-Bogard and Davis 2014, Ludlow et al. 2015). Such changes

could increase the risk of brood parasitism and predation, as well as lowering foraging

efficiency, leading to reduced survival rates.

The quality of vegetation in grassland habitat may also influence grassland

songbird abundance (Davis et al. 2013). Kalyn-Bogard and Davis (2014) postulated that

good management decisions that lead to healthy vegetation structure in rangeland might

mitigate negative effects of energy development on grassland bird abundance. Thus,

there may be interactive effects between the quality of grassland vegetation, grass type

and oil wells on bird abundance. For example, bird abundance may be higher in

appropriately managed native pastures with oil development compared to non-native

pastures.

If oil development influences predator behaviour this could also have impacts on

grassland songbird abundance and reproductive success. Grassland songbird

reproductive success is largely driven by predation on eggs and chicks at nest sites

(Martin 1988, Koford 1999, Davis 2003, Ribic et al. 2012), with grassland-nesting birds

experiencing higher predation rates than above-ground nesting species (Ricklefs 1969,

Martin 1993). Much speculation exists regarding the effect that oil development might

have on the occurrence and activity of grassland songbird predators (Ingelfinger and

Anderson 2004, Kalyn-Bogard and Davis 2014). For example, raptors and corvids are

known to perch and nest on power poles and power lines (Knight and Kawashima 1993,

5

Ingelfinger and Anderson 2004). Such vertical structures may attract more avian

predators and provide them with places to actively hunt from in an otherwise flat

landscape. However, there is little published research on the behaviour and activity of

grassland songbird predators in response to oil development. Understanding the extent to

which predators are influenced by oil development will allow researchers and land

managers to identify possible mechanisms affecting grassland songbird demography in

areas where oil extraction occurs.

Environment Canada has issued standardized guidelines for petroleum industry

activities that affect threatened and endangered species, including grassland birds

(Environment Canada 2009). These guidelines regulate the amount and proximity of

industrial activity in locations where species at risk occur. This includes temporal and

spatial restrictions (e.g. setback distances from active breeding and nesting sites, no

development during breeding seasons). Consideration of the cumulative effects of

disturbance features is also suggested (Environment Canada 2009), yet there are no

guidelines to enforce or determine a cumulative disturbance cutoff. Due to a lack of

research on the effects of oil development in grassland habitat, most recommendations

are based on expert knowledge instead of empirical evidence. There is a knowledge gap

when it comes to assessing ongoing and cumulative effects of energy development on

grassland species and habitat (Nasen et al. 2011). Such information is needed to

effectively assess the impacts of oil development on grassland songbirds to determine

whether existing largely qualitative guidelines are reasonable.

The purpose of my research was to determine the effects of oil development on

grassland songbirds. I used data collected from point counts and vegetation surveys to

6

determine the influence of oil well proximity, density, and associated disturbance features

on grassland songbird abundance. I also used a subset of these same data to determine

the effects of oil well density and associated disturbance features on the occurrence of

grassland avian predators. My research aims to provide land managers and responsible

government departments with empirical information on the extent to which oil

development impacts grassland songbirds.

7

1.2 LITERATURE CITED

Askins, R. A., F. Chávez-Ramírez, B. C. Dale, C. A. Haas, J. R. Herkert, F. L. Knopf,

and P. D. Vickery. 2007. Conservation of grassland birds in North America:

understanding ecological processes in different regions: “Report of the AOU

Committee on Conservation”. Ornithological Monographs:iii–46.

Bayne, E. M., and B. C. Dale. 2011. Effects of energy development on songbirds. Pages

95–114 in D. E. Naugle, editor. Energy Development and Wildlife Conservation in

Western North America. Island Press/Center for Resource Economics.

Bayne, E. M., L. Habib, and S. Boutin. 2008. Impacts of chronic anthropogenic noise

from energy-sector activity on abundance of songbirds in the Boreal Forest.

Conservation Biology 22:1186–1193.

Bayne, E. M., S. L. Van Wilgenburg, S. Boutin, and K. A. Hobson. 2005. Modeling and

field-testing of Ovenbird (Seiurus aurocapillus) responses to boreal forest dissection

by energy sector development at multiple spatial scales. Landscape Ecology 20:203–

216.

Brennan, L. A., and W. P. Kuvlesky. 2005. North American grassland birds: an unfolding

conservation crisis? The Journal of Wildlife Management 69:1–13.

Dale, B. C., T. S. Wiens, and L. E. Hamilton. 2009. Abundance of three grassland

songbirds in an area of natural gas infill drilling in Alberta, Canada. Proceedings of

the Fourth International Partners in Flight Conference: Tundra to Tropics:194–204.

Davis, S. K. 2003. Nesting ecology of mixed-grass prairie songbirds in southern

Saskatchewan. The Wilson Bulletin 115:119–130.

8

Davis, S. K. 2004. Area sensitivity in grassland passerines: effects of patch size,

patch shape, and vegetation structure on bird abundance and occurrence in southern

Saskatchewan. Auk 121:1130–1145.

Davis, S. K., R. J. Fisher, S. L. Skinner, T. L. Shaffer, and R. M. Brigham. 2013.

Songbird abundance in native and planted grassland varies with type and amount of

grassland in the surrounding landscape. The Journal of Wildlife Management

77:908–919.

Environment Canada. 2009. Petroleum industry activity guidelines for wildlife species at

risk in the Prairie and Northern region. Canadian Wildlife Service, Environment

Canada, Prairie and Northern Region, Edmonton Alberta. 64p.

Fisher, R. J., and S. K. Davis. 2011. Post-fledging dispersal, habitat use, and survival of

Sprague’s Pipits: are planted grasslands a good substitute for native? Biological

Conservation 144:263-271.

Gilbert, M. M., and A. D. Chalfoun. 2011. Energy development affects populations of

sagebrush songbirds in Wyoming. Journal of Wildlife Management 75:816–824.

Government of Saskatchewan: Economy 2015. Oil and gas. Available at:

http://www.economy.gov.sk.ca/oilandgas Accessed June 25, 2015.

Habib, L., E. M. Bayne, and S. Boutin. 2007. Chronic industrial noise affects pairing

success and age structure of Ovenbirds Seiurus aurocapilla. Journal of Applied

Ecology 44:176–184.

Hammermeister, A., D. Gauthier, and K. McGovern. 2001. Saskatchewan’s native

prairie: taking stock of a vanishing ecosystem and dwindling resource. Saskatoon,

Saskatchewan: Native Plant Society of Saskatchewan Inc.

9

Henderson, A. E., and S. K. Davis. 2014. Rangeland health assessment: a useful tool for

linking range management and grassland bird conservation? Rangeland Ecology &

Management 67:88-98.

Hickman, K. R., G. H. Farley, R. Channell, and J. E. Steier. 2006. Effects of old world

bluestem (Bothriochloa ischaemum) on food availability and avian community

composition within the mixed-grass prairie. The Southwestern Naturalist 51:524–

530.

Ingelfinger, F., and S. Anderson. 2004. Passerine response to roads associated with

natural gas extraction in a sagebrush steppe habitat. Western North American

Naturalist 64:385–395.

Kalyn-Bogard, H. J., and S. K. Davis. 2014. Grassland songbirds exhibit variable

responses to the proximity and density of natural gas wells. The Journal of Wildlife

Management 78:471–482.

Knight, R. L., and J. Y. Kawashima. 1993. Responses of raven and Red-tailed Hawk

populations to linear right-of-ways. The Journal of Wildlife Management 57:266–

271.

Koford, R. R. 1999. Density and fledging success of grassland birds in Conservation

Reserve Program fields in North Dakota and west-central Minnesota. Studies in

Avian Biology 19:187-195.

Lawson, A. L., M. L. Morrison, and R. D. Slack. 2011. Impacts of oil and gas

development on wintering grassland birds at Padre Island National Seashore, Texas.

Southeastern Naturalist 10:303–320.

10

Leddy, K. L., K. F. Higgins, and D. E. Naugle. 1999. Effects of wind turbines on upland

nesting birds in Conservation Reserve Program grasslands. The Wilson Bulletin

111:100–104.

Linnen, C. G. 2008. Effects of oil and gas development on grassland birds. Unpublished

report, prepared for Petroleum Technology Alliance Canada. Saskatoon,

Saskatchewan, Canada.

Ludlow, S. M., R. M. Brigham, and S. K. Davis. 2015. Oil and natural gas development

has mixed effects on the density and reproductive success of grassland songbirds.

The Condor 117:64–75.

Martin, T. E. 1988. Processes organizing open-nesting bird assemblages: competition or

nest predation? Evolutionary Ecology 2:37-50.

Martin, T. E. 1993. Nest predation among vegetation layers and habitat types: revising

the dogmas. The American Naturalist 141:897–913.

Murphy, M. T., and F. Moore. 2003. Avian population trends within the evolving

agricultural landscape of eastern and central United States. The Auk 120:20–34.

Nasen, L. C., B. F. Noble, and J. F. Johnstone. 2011. Environmental effects of oil and gas

lease sites in a grassland ecosystem. Journal of Environmental Management 92:195–

204.

North American Bird Conservation Initiative Canada. 2012. The state of Canada’s birds,

2012. Environment Canada, Ottawa, Canada. 36p.

North American Bird Conservation Initiative U.S. Committee. 2009. The state of the

birds, United States of America, 2009. U.S. Department of Interior: Washington,

DC. 36p.

11

Ribic, C. A., M. Guzy, T. Anderson, D. Sample, and J. Nack. 2012. Bird productivity and

nest predation in agricultural grasslands. Video Surveillance of Nesting Birds.

Studies in Avian Biology No. 43. Cooper Ornithological Society. 240p.

Ribic, C. A., R. R. Koford, J. R. Herkert, D. H. Johnson, N. D. Niemuth, D. E. Naugle, 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.

Ricklefs, R. E. 1969. Nesting mortality in birds. Smithsonian Contributions to Zoology

No. 9. 56p.

Samson, F., and F. Knopf. 1994. Prairie conservation in North America. BioScience

44:418–421.

Sauer, J. R., J. E. Hines, J. E. Fallon, K. L. Pardieck, D. J. Ziolkowski, Jr., and W. A.

Link. 2014. The North American breeding bird survey, results and analysis 1966 -

2013. Version 01.30.2015 USGS Patuxent Wildlife Research Center, Laurel, MD.

Smallwood, K. S. 2013. Comparing bird and bat fatality-rate estimates among North

American wind-energy projects. Wildlife Society Bulletin 37:19–33.

Souther, S., Tingley, M. W., Popescu, V. D., Hayman, D. T. S., Ryan, M. E., Graves, T.

A., Hartl, B., and Terrell, K. 2014. Biotic impacts of energy development from

shale: research priorities and knowledge gaps. Frontiers in Ecology and the

Environment 12:330-338.

Stevens, T. K., A. M. Hale, K. B. Karsten, and V. J. Bennett. 2013. An analysis of

displacement from wind turbines in a wintering grassland bird community.

Biodiversity and Conservation 22:1755–1767.

12

2. EFFECTS OF OIL WELL PROXIMITY, DENSITY AND OVERALL

DISTURBANCE ON GRASSLAND SONGBIRD ABUNDANCE

2.1 INTRODUCTION

As the human population increases, the need to exploit natural resources leads to

increased habitat degradation and destruction, and eventually the extinction of individual

species (Gaston et al 2003, Leu et al. 2008). Loss and degradation of global ecosystems

has largely been attributed to such anthropogenic activities as agriculture, forestry,

mining, and energy development. Grasslands have been drastically altered by human

development and are considered one of the most altered and threatened ecosystems in

North America (Noss et al. 1995, Noss 2000, Askins et al. 2007).

The major causes associated with loss and degradation of grasslands are

conversion to cropland, afforestation, fragmentation, human settlement, fire suppression,

exotic species and woody vegetation encroachment, and deterioration of rangelands

(Murphy and Moore 2003, Brennan and Kuvlesky 2005, Askins et al. 2007). However,

oil development may also be threatening the quality and quantity of the remaining

grassland habitat in North America. Oil extraction is a common activity in North

American grasslands and is rapidly expanding. Saskatchewan alone had an estimated

30,756 actively producing oil wells in 2014 (Government of Saskatchewan 2015), with

thousands of new wells being drilled every year. The potential impact of oil development

on grassland ecosystems include linear (roads, pipelines, power lines, and fences) and

non-linear (wells, buildings, soil compaction, noise, traffic, and invasive species)

disturbances that may individually or collectively have negative impacts on grassland

habitats and consequently species density, survival and reproductive success. Wildlife

13

may avoid areas near oil wells because of increased noise and activity (Harju et al. 2010),

or because it represents poor quality habitat (Sawyer et al. 2009, Beckmann et al. 2012),

which reduces survival and reproduction (Northrup and Wittemyer 2013). These

responses may also drive landscape-level effects, whereby species demographic rates

may be lower in areas disturbed by oil development than areas either without, or with

little, disturbance.

Grassland songbirds are one group of grassland wildlife that may be particularly

negatively influenced by oil development. Grassland birds have exhibited steeper

declines than any other group of birds in North America in recent decades; 22 of 28

obligate grassland species for which there are sufficient data are declining (Sauer et al.

2014). As a result, declines of North American grassland birds are emerging as a

prominent conservation crisis of the 21st century (Brennan and Kuvlesky 2005).

Increasing oil development on grasslands may be contributing to the observed population

declines of grassland songbirds.

Few studies have examined the effects of oil development on grassland songbirds,

and the general response of bird communities to industrial development is poorly

understood (Bayne and Dale 2011, Northrup and Wittemyer 2013, Souther et al. 2014).

Grassland birds may avoid areas near oil wells because they experience reduced survival

and reproduction (Lyon and Anderson 2003, Aldridge and Boyce 2007, Gilbert and

Chalfound 2011, Ludlow et al. 2015). Subsequently, grassland bird density may be lower

in and around sites associated with energy development and extraction because these

areas represent poor quality habitat (Dale et al. 2009, Lawson et al. 2011, Kalyn-Bogard

and Davis 2014). Grassland songbirds may also avoid oil infrastructure because of

14

increased noise (Bayne et al. 2008, Francis et al. 2009, Bayne and Dale 2011), human

activity and traffic (Ingelfinger and Anderson 2004), or invasion by non-native plant

species (Ludlow et al. 2015). Songbirds also likely experience direct mortality and nest

loss from seismic exploration, vehicle traffic, and clearing of grassland for pipelines and

well pads (Van Wilgenburg et al. 2013). The amount of disturbance (e.g. roads, pipelines,

exotic vegetation, buildings and batteries) associated with oil extraction may also

negatively influence the abundance or reproductive success of grassland songbirds.

Many grassland birds are area sensitive (abundance declines with reduced patch sizes;

Davis 2004, Ribic et al. 2009) and have lower reproductive success in fragmented

landscapes (Herkert et al. 2003, Davis et al. 2006). Thus, grassland birds may experience

reduced abundance due to edge effects and habitat fragmentation caused by oil

development (Ingelfinger and Anderson 2004, Bayne et al. 2005, Koper et al. 2009).

Grassland songbirds may be sensitive to the cumulative amount of disturbance in a

landscape as well as single disturbance structures, although cumulative effects on most

species are poorly understood currently (Bayne and Dale 2011, Souther et al. 2014).

While the exact mechanisms are unknown, avoidance of oil wells and associated

disturbance features may result from changes in vegetation structure surrounding these

sites (Dale et al. 2009). Similar to natural gas development, oil development could

increase invasive exotic grasses (Kalyn-Bogard and Davis 2014, Ludlow et al. 2015), or

increase soil compaction due to greater cattle activity around oil wells and development

features (Kalyn-Bogard and Davis 2014). There may also be edge effects on vegetation

structure associated with well pads and roads that influence grassland songbird

abundance (Koper et al. 2009, Sliwinski and Koper 2012). Such changes could increase

15

the risk of brood parasitism or predation to nests and individuals, lower foraging

efficiency, or reduce survival rates of adults and juveniles.

The amount and type of grassland may also influence the abundance of grassland

songbirds. Grassland specialist abundance is greater on native pastures (Davis et al.

2013), while non-native patches of grassland support a lower abundance of birds (Dale et

al. 1997, Flanders et al. 2006), lower diversity (Giuliano and Daves 2002, George et al.

2013), and lower reproductive success (Lloyd and Martin 2005, Fisher and Davis 2011).

Furthermore, grassland specialists are more common in planted and native grass patches

surrounded by native grasslands (Davis et al. 2013). Agriculture and Agri-Food

Canada’s Permanent Cover Program (PCP) converted cropland to hayland and pasture

(McMaster and Davis 2001; McMaster et al. 2005), but past government policies

encouraged the planting of exotic species for tame forage and hay production (Sutter and

Brigham 1998). Likewise, the Food Security Act in the United States established the

Conservation Reserve Program (CRP) in 1985 to conserve and improve soil and water

resources and limit erosion of cropland through establishing perennial cover (e.g. grasses

and legumes; Johnson and Schwartz 1993). These planted grasslands are beneficial to

landowners because they are low cost and have rapid growth, but they do not restore

native prairie or their avian communities (Johnson and Schwartz 1993; McMaster and

Davis 2001). Oil development occurs on both native and planted grasslands, and while

habitat quality may be reduced by oil development in both habitat types, the effect may

be stronger in planted grasslands if they are of lesser quality. Species known to prefer

native grass pastures (grassland specialists) would be expected to exhibit larger negative

16

effects in relation to well proximity, density, and cumulative disturbance in planted

pastures compared to grassland generalists.

Environment Canada has issued standardized guidelines for petroleum industry

activities that affect threatened and endangered species, including grassland birds

(Environment Canada 2009). However, due to a lack of quantitative research specifically

assessing the impacts of oil development on grassland birds, these guidelines were based

primarily on expert qualitative opinion. More studies are needed to effectively assess the

impacts of oil development on wildlife to determine whether existing guidelines are

reasonable. Therefore, the purpose of my research was to determine the extent to which

oil development affects grassland songbird abundance to better inform conservation

decisions.

I hypothesized that the presence of oil wells and associated disturbances created

by oil development would reduce the amount of suitable habitat for grassland songbirds.

Furthermore, I hypothesized that planted grassland is lower quality habitat than native

grassland for grassland specialists (e.g. Baird’s Sparrow [Ammodramus bairdii],

Chestnut-collared Longspur [Calcarius ornatus], Sprague’s Pipit [Anthus spragueii]) but

not grassland generalists (e.g. Brown-headed Cowbird [Molothrus ater], Savannah

Sparrow [Passerculus sandwichensis], Vesper Sparrow [Pooecetes gramineus]).

Therefore I predicted that for grassland specialists, abundance would be lower in planted

grassland and that the negative effect of well distance and density and cumulative

disturbance would be greatest in planted grassland than native grassland. I predicted that

the abundance of grassland generalists would be similar in native and planted grassland

and that the effect of well distance and density and cumulative disturbance would be the

17

same in native and planted grassland. However, since Bobolinks (Dolichonyx

oryzivorus) are typically more common in planted compared to native pastures (Davis et

al. 2013), I predicted that Bobolink abundance would be greater in planted grassland and

that the negative effect of well distance and density and cumulative disturbance would be

greatest in native compared to planted grassland. Furthermore, I hypothesized that

cumulative disturbance would reduce the amount of suitable habitat more than any single

disturbance feature associated with oil development. Thus, I predicted that cumulative

disturbance would have a larger negative effect on abundance than any single disturbance

feature.

2.2 METHODS

2.2.1 Study area

I conducted my study in southeastern Saskatchewan (Fig. 2.1) in 2013 and 2014,

in an area of active oil development and extraction along the edge of the mixed-moist

grasslands and aspen parkland ecoregions. The region is characterized by rolling fescue

grasslands dotted with aspen bluffs and wetlands. Most of the region is cultivated to

produce a wide variety of cereals, oil seeds, feed grains, and forage crops. Native

vegetation is primarily confined to non-arable pasture-lands, and often borders rivers and

streams. Native vegetation consists primarily of aspen (Populus tremuloides) bluffs,

shrubs such as western snowberry (Symphoricarpos occidentalis), prairie rose (Rosa

arkansana), chokecherry (Prunus virginiana), and wolf willow (Elaeagnus commutata),

and a mix of speargrasses (Aristida spp.) and wheatgrasses (Agropyron spp.). A large

portion of the land was seeded to exotic grasses and legumes for pasture and hay.

18

Planted pastures are typically composed of crested wheatgrass (Agropyron cristatum),

smooth brome (Bromus inermis), and Kentucky bluegrass (Poa pratensis) and hayfields

are typically characterized by smooth brome grass and alfalfa (Medicago spp.). The area

has experienced increased oil development since 2008, when hydraulic fracturing was

introduced (Government of Saskatchewan 2015). As a result, oil wells in the study area

are a mixture of conventional extraction and hydraulic fracturing.

2.2.2 Study site selection

I randomly selected oil wells across a gradient of oil development, in native and

planted pastures using a Geographic Information System (ArcGIS 10.0-10.2, ESRI). I

buffered each well with a 908 m radius (2.59 km2) buffer and quantified the density of oil

wells in each buffer. A well density gradient was created based on four density

categories: none, low (1-4 wells), medium (5-8), and high (≥9). These categories were

chosen based on well density categories used in similar studies (Dale et al. 2009;

Hamilton et al. 2011). I refer to the buffered areas as my sample sites and the well at the

center of each buffer as my focal well for each sample site. Well density was calculated

as the number of wells/259 ha (908 m radius) because this represents the size of a section

of land (McKercher and Wolfe 1986) about which management decisions are often made.

I ensured that sample sites did not overlap each other. Focal wells and sample sites were

randomly selected, however, sample site locations were ultimately subject to acquiring

permission from landowners.

Oil wells were divided into ‘active’ and ‘abandoned’ categories. An active well

consisted of a productive well with a pump-jack and gravel well pad. The size of well

pads ranged from traditional tear-drop shaped gravel pads to pads that had been stripped

19

of all top soil and piled in burms around the entire lease site. Also, some wells had oil

storage tanks located on the pad and some sites included more than one well on the same

pad. Pump-jacks at active wells were driven by relatively quiet electric motors. An

operator visited all active well sites on a daily basis. Abandoned wells were non-

productive wells that ranged from a capped pipe with vegetation growing up to the pipe

to sites with minimal gravel pads and old pump-jack structures left in place. No visits.

Two bird survey locations were randomly associated with each focal well: one

within a 100 m buffer of the well (near), and the second within a 100-400 m buffer of the

well (far) (Fig. 2.2) to ensure I captured a gradient of well distances. Near and far survey

locations were situated >200 m from each other. Grassland habitat associated with each

well location was verified through ground-truthing following bird surveys. Habitat type

(‘native’ vs. ‘planted’) was determined by the dominant vegetation type of the quarter

section (McKercher and Wolfe 1986) that each bird survey was located in.

2.2.3 Avian and vegetation surveys

I quantified bird abundance using five-minute point-count surveys. Each point-

count location was surveyed four times by four different observers within two days of

each other to ensure that I was sampling a closed population (i.e. birds were not

emigrating or immigrating from the point-count location within the sampling period).

Repeated visits by different observers allowed me to account for birds present but

undetected and estimate detection probabilities for each species, allowing for robust

abundance estimates (MacKenzie et al. 2002, MacKenzie and Royle 2005, Johnson 2008).

Bird surveys were conducted during the breeding season after breeding territories had

been established (May 20 to July 7, 2013 and 2014). No surveys were conducted after

20

July 7 to ensure that only local breeders for that year were included in abundance

estimates. Survey locations were rotated every second day between the western, middle

and eastern portions of the study area. All point-counts were conducted between sunrise

and 1000 hrs, when winds were <20 km/h, and there was little to no precipitation.

Observers recorded the species of all birds seen or heard within 100 m of the point count

center.

Vegetation structure was assessed at each point-count location to help determine

whether variation in abundance was a result of factors related to oil development or

habitat. Vegetation surveys were conducted on the same days as avian surveys, and were

randomly located within 1-100 m of the point count in each of the four cardinal directions.

A 50 cm x 50 cm quadrat was randomly placed at each vegetation survey location and the

percent cover was recorded of native grass (live and dead combined), exotic grass (live

and dead combined), forbs, shrubs, bare ground (exposed soil, club moss, lichen, and

abiotic material). Vegetation height (80% of the maximum standing, attached vegetation;

cm) and litter depth (dead, detached vegetation; mm) were measured in the center of the

quadrat using a ruler. Distance to the nearest shrub (m) and visual obstruction (measured

using a Robel pole; Robel et al. 1970) were also measured from the center of the quadrat.

2.2.4 Disturbance measurements

I located all accessible roadways, pipelines, well sites, power lines, fences, oil

batteries, and other buildings and disturbance features associated with oil development

with a hand-held Global Positioning System (GPS) unit and converted the locations to

point, line and polygon features in ArcMap (ArcGIS 10.0-10.2, ESRI). I also used colour

digital ortho-photos taken during 2008-2012 at a 0.6 m resolution and a 1:1000 scale,

21

acquired from the Saskatchewan Geospatial Imagery Collaborative (Saskatchewan

Research Council 2015), to quantify any disturbance features that were not accessible on

the ground. Linear features (roads and trails, pipelines, and grid roads) were buffered

based on the average width of 36 randomly selected oil roads/trails, grid roads, and

pipelines respectively. The areas of all well pads, battery and building pads, oil roads and

trails, pipelines, grid roads, and the length of power lines were summed for each sample

site. A cumulative disturbance category was also created by summing the area

encompassed by all well pads, battery and building pads, oil roads and trails, and

pipelines for each sample site.

2.2.5 Statistical analysis

I performed all analyses in R (v. 3.1.1. “Sock it to Me”, The R Foundation for

Statistical Computing, 2014).

2.2.5.1 Vegetation and oil well models

I assessed correlations between all vegetation variables to determine if any were

highly correlated with one another (r>0.65). Only visual obstruction and vegetation

height were positively correlated (r=0.69). Therefore, I used the variable with the smaller

Akaike’s Information Criterion (AIC) value for each species in the vegetation models

(see below). I then assessed how variation in vegetation structure changed in relationship

to proximity and density of oil wells by comparing seven models for each vegetation

variable (well density, well distance, well density + well distance, well distance x habitat

[native or planted], well density x habitat, well distance x habitat + well density x habitat,

and a null model). I ranked each model using AIC values and weights (Burnham and

Anderson 1998). I used 85% confidence limits to identify uninformative parameters

22

(Arnold 2010) and to determine the strength of effect for each informative parameter.

Uninformative parameters in the top abundance models are not discussed. I considered

the top fitted model to be the one with the highest AIC weight (wi). If there were

competing models (∆AICc < 2 and the same or fewer parameters as the model with the

highest AIC weight), I selected the most parsimonious model as the top model. I used

this method for model selection in all analyses.

2.2.5.2 Bird abundance models

I used the ‘pcount’ function in the “unmarked” package (Fiske and Chandler

2011) in R for all bird abundance analyses. The ‘pcount’ function uses hierarchical N-

mixture models developed by Royle (2004) to estimate abundance from spatially

replicated count data. N-mixture models have two ‘sides’: an observation model, which

uses a Binomial distribution to estimate a detection probability, and an abundance model,

which is directly informed by the observation model and uses a Poisson distribution to

estimate the site-specific abundance (‘N’).

I first fit the best observation model for each bird species using my count data and

four detection parameters (ordinal date of survey, time of survey, wind speed during

survey, and observer). I also included two site-specific parameters in the observation

models (well density and well distance) in case the presence, movement, or noise from oil

wells influenced bird detection. I used abundance counts of singing males within a 100

m radius for all species, except Brown-headed Cowbirds where I used the total counts of

both males and females. I compared the interactive and additive effects of year for each

detection parameter to determine if I could combine the data from both years. I then

created a global model by comparing each detection parameter to the Null model and

23

retained the detection parameters that performed better than the Null model. I fit all

subset models of the global model and selected the top observation model to be used as

the detection component of the abundance model for each species (Appendix A).

To avoid over-parametizing my models, I determined the relative influence of

vegetation structure and oil wells on songbird abundance by taking a step-wise approach

to model selection. I first identified the top vegetation and oil development models for

each species and then compared their AICc values with 1) each other, 2) a model

composed of the top vegetation and oil well models, 3) a model composed of the top

vegetation and oil well models with the additive or interactive effects of habitat type

(native or planted), 4) a detection-only model, and 5) a null model. I started by

determining the best vegetation model for each species. I considered eight vegetation

parameters: native grass cover, exotic grass cover, forb cover, bare ground cover,

vegetation height, distance to nearest shrub, litter depth, and visual obstruction. I first

evaluated the interactive and additive effects of year with each vegetation parameter to

determine if I could combine the data from both years. I then compared the linear and

quadratic effect of each vegetation parameter to the observation only model and selected

the linear or quadratic relationships that performed better than the observation model for

each species. I fit all subsets of the remaining vegetation parameters and selected the top

vegetation model for each species (Appendix B).

I identified the best oil well model for each species by considering three oil-well

variables: distance to nearest well, well density, and well activity. Well activity was

categorically assessed by whether the focal well was an actively producing well (‘active’),

a well no long producing oil (‘abandoned’), or no well present (‘none’). I used the same

24

approach as the vegetation models to evaluate the effects of year and to select appropriate

oil well variables for the global oil well models. I fit all model subsets of the global oil

well model and selected the top oil well model for each species (Appendix C). Any

uninformative parameters (85% CI includes zero) included in the top abundance models

are not discussed.

2.2.5.3 Landscape disturbance models

I assessed five landscape-scale disturbance features associated with oil

development to determine their effect on bird abundance. The landscape scale features

included the area of: pipelines, battery and oil building pads, well pads, oil roads/trails,

and cumulative disturbance (sum of all disturbance features). I limited the disturbance

analysis to the subset of point-counts closest to the focal well of each sample area. This

was to ensure that disturbance analysis was related to bird abundance from the point-

count closest to the center of the sample area.

To select the top abundance model associated with landscape disturbance for each

species I compared the AICc values of 1) a model composed of the top disturbance

features associated with oil development, 2) a model composed of the top oil

development disturbance features with the additive and interactive effects of habitat type

(native vs. planted), 3) a detection-only model, and 4) a null model. I first assessed

correlations between all disturbance features to determine if any disturbance types were

highly correlated with one another. I also evaluated the interactive and additive effects of

year with each disturbance parameter to determine if I could combine the two years of

data. I then compared the effect of each disturbance parameter on bird abundance to the

detection model and selected the disturbance parameters that performed better than the

25

detection model for each species. I fitted all subsets of the remaining disturbance

parameters and selected the top disturbance model for each species (Appendix D).

2.3 RESULTS

2.3.1 General Results

I conducted 486 point counts at 243 sample sites (112 native and 131 planted)

across southeastern Saskatchewan in 2013-2014 (Table 2.1). The counts were distributed

across a gradient of no well (55), low density (62), medium density (61) and high well

density (65) sample sites (Table 2.1).

Eleven species of grassland songbird were detected at >10% of my sample sites

and were included in my analyses. The most common species (≥70% frequency of

occurrence) were Savannah Sparrow, Brown-headed Cowbird, Clay-colored Sparrow

(Spizella pallida), and Western Meadowlark (Sturnella neglecta; Table. 2.2). Vesper

Sparrow, Bobolink, Grasshopper Sparrow (Ammodramus savannarum), and Baird’s

Sparrow were recorded at 35-50% of point-counts (Table 2.2). Sprague’s Pipit, Horned

Lark (Eremophila alpestris), and Chestnut-collared Longspur had the lowest occurrence

frequencies (>10 but <25%; Table 2.2).

2.3.2 Vegetation

All vegetation variables differed significantly between native and planted pastures,

except litter depth (Table 2.3). Vegetation height, visual obstruction, forb cover, and

exotic grass cover were all greater in planted pastures (Table 2.3). Distance to shrubs and

percent native cover was greatest in native pastures (Table 2.3).

26

Vegetation structure varied with well proximity and well density except for forb

cover and litter depth (Table 2.4). Bare ground decreased with increasing distance to a

well (Fig. 2.3). Visual obstruction increased with increased distance to a well and with

increased well density (Fig. 2.6). Relationships between native grass cover, exotic grass

cover, and shrub distance varied with well density or well distance and depended on

habitat type (Table 2.4). The proportion of native cover increased with increased

distance to a well in native pastures, but the confidence limits suggest no effect in planted

pastures (Fig. 2.4). However, the proportion of exotic grass cover decreased with

increasing distance from a well in native pastures, but there was no effect in planted

pastures (Fig. 2.5). Shrub distance was greater in planted pastures and decreased with

increased distance to a well and increased well density in both habitat types (Fig. 2.8);

however, the confidence limits suggest that the effect was small.

2.3.3. Effects of well proximity, well density and vegetation on bird abundance

The vegetation models explained variation in abundance better than the oil-well

models alone for all species except Grasshopper Sparrow. Grasshopper Sparrow

abundance exhibited an interactive effect between year and litter depth, with opposite

trends in 2013 and 2014 (Appendix B). Therefore I analyzed the data for 2013 and 2014

separately. For Baird’s Sparrow, Chestnut-collared Longspur and Sprague’s Pipit, the

vegetation model was ≥70 AIC units smaller than the oil-well model, and 3-25 AIC units

smaller for all other species (Table 2.6). Nine grassland songbird species had some

combination of vegetation and oil-well parameters included in the top abundance model

(Table 2.5). For these species, the combination of vegetation and oil-well features better

explained variation in abundance than either of these factors alone.

27

Abundance of eight species was influenced by visual obstruction or vegetation

height (Table 2.5). Abundance of Western Meadowlark (Fig. 2.19), Sprague’s Pipit (Fig.

2.18), Vesper Sparrow (Fig. 2.17), and Horned Lark (Fig. 2.21) decreased as visual

obstruction increased. Clay-colored Sparrow (Fig. 2.20) and Brown-headed Cowbird

(Fig. 2.11) abundance increased as visual obstruction increased and Bobolink abundance

increased as vegetation height increased (Fig. 2.12). In 2013, Grasshopper Sparrow

abundance was greatest with a moderate amount of visual obstruction (Fig. 2.14).

The abundance of seven species were affected by litter depth (Table 2.5).

Chestnut-collared Longspur (Fig. 2.13), Sprague’s Pipit (Fig. 2.18), Baird’s Sparrow (Fig.

2.10), Horned Lark (Fig. 2.21), Western Meadowlark (Fig. 2.19), and Grasshopper

Sparrow (2014 only; Fig. 2.15) abundance decreased as litter depth increased. Only

Clay-colored Sparrow abundance increased with greater litter depth (Fig. 2.20).

The abundance of five species depended on whether cover was native or exotic

(Table 2.5). Baird’s Sparrow (Fig. 2.10) and Sprague’s Pipit (Fig. 2.18) abundance

increased as native grass cover increased. Chestnut-collared Longspur abundance

decreased with more exotic grass cover (Fig. 2.13), whereas Bobolink abundance

increased as exotic grass cover increased (Fig. 2.12). Vesper Sparrow abundance was

greatest when there was a mix of native and exotic grass cover (Fig. 2.17). Furthermore,

Chestnut-collared Longspur (Fig. 2.13), Horned Lark (Fig. 2.21), and Sprague’s Pipit

(Fig. 2.18) exhibited greater abundance in native pastures compared to planted pastures,

but Bobolink abundance was greater in planted pastures (Fig. 2.12).

The abundance of four species was affected by distance to the nearest shrub

(Table 2.5). Bobolink (Fig. 2.12), Savannah Sparrow (Fig. 2.16), and Grasshopper

28

Sparrow (2014 only; Fig. 2.15) abundance increased further from shrubs, whereas Clay-

colored Sparrow abundance increased as shrub distance decreased (Fig. 2.20). Also,

Baird’s Sparrow (Fig. 2.10) and Chestnut-collared Longspur (Fig. 2.13) abundance

decreased as forb cover increased, while Savannah Sparrow abundance decreased as the

amount of bare ground increased (Fig. 2.16).

Abundance of three species was influenced by an interactive effect of habitat type

and well proximity or activity (Table 2.5). Baird’s Sparrow abundance increased with

increased distance to a well, but the effect was stronger near active wells compared to

abandoned wells (Fig. 2.9). Baird’s Sparrow was also most abundant in native pastures

compared to planted pastures when a well was present, but there was no difference in

abundance when no well was present (Fig. 2.10). Bobolink abundance also increased as

well distance increased, but this effect was greater in native pastures (Fig. 2.12). Brown-

headed Cowbird abundance was greatest near abandoned well sites in native pastures (Fig.

2.11).

The abundance of five species was influenced by the main effect of well

proximity or density (Table 2.5). Chestnut-collared Longspur (Fig. 2.13) and Savannah

Sparrow abundance (Fig. 2.16) decreased closer to wells, whereas Vesper Sparrow

abundance increased closer to wells (Fig. 2.17). Sprague’s Pipit (Fig. 2.18) and Western

Meadowlark (Fig. 2.19) abundance decreased as well density increased beyond 7

wells/259 ha.

In 2013, Grasshopper Sparrow abundance was influenced by a quadratic effect of

well proximity (Table 2.5) and was greatest at ~450 m from a well (Fig. 2.14). In 2014,

29

Grasshopper Sparrow abundance was greatest in areas with active wells, but abundance

increased further from wells (Fig. 2.15).

2.3.4. Cumulative disturbance effects on bird abundance

All types of oil disturbance features increased as well density increased (Table

2.7), and subsequently cumulative disturbance also increased (Fig. 2.22). There was no

difference between the amount of disturbance in native and planted sites (Table 2.7).

There was a positive correlation (r>0.70) between well density and well pads and oil

roads (Table 2.8); however, no other disturbance features were strongly correlated

(r<0.54).

Four species were influenced by the amount of cumulative disturbance associated

with oil development and three species were influenced by single disturbance features

(Table 2.9). Bobolink (Fig. 2.23) and Savannah Sparrow (Fig. 2.23) abundance

decreased as cumulative disturbance increased. However, Grasshopper Sparrow

abundance was greatest when cumulative disturbance was ~3% of the total landscape

(Fig. 2.23), whereas Clay-colored Sparrow abundance was lowest at the same amount of

cumulative disturbance (Fig. 2.23). Abundance of Brown-headed Cowbird, Horned Lark,

Sprague’s Pipit, and Vesper Sparrow was not influenced by the amount of disturbance in

the surrounding landscape (Table 2.9).

Grasshopper Sparrow abundance increased as the amount of area disturbed by oil

battery and building pads increased (Fig. 2.25). In contrast, Baird’s Sparrow abundance

decreased as well pad area increased (Fig. 2.24). Western Meadowlark abundance was

most influenced by the interactive effect between oil roads and habitat type (Table 2.9).

30

Western Meadowlark abundance was greatest when oil road area was ~0.5%, and then

declined as oil road area increased, but only in planted pastures (Fig. 2.26).

2.4. DISCUSSION

My results showed that vegetation structure was better at explaining variation in

abundance than distance to oil wells, well density, and well type (active or abandoned)

for all but one grassland songbird. However, combining vegetation structure and oil well

variables improved model fit for all species except Horned Lark and Clay-colored

Sparrow whose abundances were best explained by vegetation structure alone. Baird’s

Sparrow, Bobolink, Chestnut-collared Longspur, Grasshopper Sparrow, Sprague’s Pipit,

Savannah Sparrow, and Western Meadowlark all exhibited reduced abundance closer to

wells or with increased well density. In contrast, Brown-headed Cowbird, Grasshopper

Sparrow (2014 only) and Vesper Sparrow exhibited greater abundance closer to wells or

in areas with a well compared to areas without wells. In addition, the influence of oil

wells on the abundance of three species depended on whether the habitat was native or

planted. Furthermore, five species experienced reduced abundance as the amount of

cumulative disturbance or single disturbance features caused by oil development

increased within the 259 ha sample sites. Taken together, the results support my

hypothesis that the presence of oil wells and associated disturbances created by oil

development reduces the amount of suitable habitat for grassland songbirds, and that

planted grasslands compound the negative effects of oil development, particularly for

native grass specialists. Also, cumulative disturbance appears to reduce the amount of

suitable habitat more than any single disturbance feature associated with oil development.

31

2.4.1. Effects of well proximity, density and activity on grassland songbird

abundance

The abundance of seven species of grassland songbirds was influenced by oil well

proximity or the activity (active or abandoned) of wells. Baird’s Sparrow, Bobolink,

Chestnut-collared Longspur, Savannah Sparrow, and Grasshopper Sparrow (2014 only)

were less abundant near oil wells. These results are consistent with Kalyn-Bogard and

Davis (2014) who found that Grasshopper Sparrow and Chestnut-collared Longspur

abundance was lower near natural gas wells, and Lawson et al. (2011) who found that

some grassland songbirds were least abundant near active oil and gas wells. Negative

associations between species abundance and energy development are not restricted to

songbirds. Similar studies have found that Greater Sage-grouse (Centrocercus

urophasianus; Holloran et al. 2015) and mule deer (Sawyer et al. 2009) also avoid active

energy development infrastructure.

Sprague’s Pipit and Western Meadowlark exhibited reduced abundance in areas

with high well density. Hamilton et al. (2011) found that Sprague’s Pipit experienced

reduced abundance in areas with greater natural gas well densities and associated

disturbance (Hamilton et al. 2011). However, Kalyn-Bogard and Davis (2014) found that

Sprague’s Pipit abundance was not influenced by natural gas well density.

Reduced abundance near wells or in high well density areas may be partially

explained by changes in vegetation structure caused by oil development. I found that

visual obstruction, distance to shrubs, bare ground cover, and native and exotic grass

cover varied with well proximity and well density. Koper et al. (2014) also found shorter

vegetation and more bare ground near natural gas wells, which they suggest was

32

explained by a combination of well construction effects and increased cattle grazing near

wells. However, I accounted for vegetation effects in my abundance models. Therefore,

other factors must also be influencing abundance associated with oil wells. Birds may

avoid oil wells or high well densities because they experience reduced reproductive

success (Lyon and Anderson 2003, Aldridge and Boyce 2007, Gaudet 2013). They may

also be avoiding other disturbances associated with oil wells such as increased traffic or

human activity (Ingelfinger and Anderson 2004, Sawyer et al. 2009, Lawson et al. 2011),

increased predator or cowbird abundance (Webb et al. 2012, Ludlow et al. 2015),

increased pollutants, toxins or invasive species associated with anthropogenic activity

(Koper et al. 2009, Mineau and Whiteside 2013, Ludlow et al. 2015), or aversion to

moving vertical structures. Grassland songbirds may also be avoiding increased

anthropogenic noise caused by oil wells (Bayne et al. 2008), though the oil wells in my

study area were typically powered by quiet electric motors.

Some grassland bird species were most abundant near wells or near active wells.

Vesper Sparrow abundance was higher near wells, and Brown-headed Cowbird

abundance was greatest in areas with abandoned wells. Ludlow et al. (2015) found that

cowbird abundance was three times higher in study plots with oil and natural gas wells,

and that Vesper Sparrows nested closer to and fledged more young from nests near access

trails. Similarly, Lawson et al. (2011) found that active oil and gas wells had slightly

higher numbers of grassland birds than abandoned wells. Oil wells and other vertical

features (e.g. fences, power poles) may provide perch sites for Vesper Sparrows and

cowbirds to display. Ludlow et al. (2015) speculated that oil wells might also provide

perches for cowbirds to locate host nests. While Baird’s Sparrow and Grasshopper

33

Sparrow (2014 only) avoided areas near wells, the abundance of both species was greater

near active compared to abandoned wells. This effect was strongest in planted grasslands

for Baird’s Sparrow. It is unclear why abundance would be greater in the presence of

active wells; however, it could suggest that Baird’s Sparrow and Grasshopper Sparrow

are not negatively affected by increased traffic associated with productive oil wells. If

Brown-headed Cowbirds and predators also avoid active wells then greater abundance

near active wells may reflect a favoured strategy.

2.4.2. Interactive effects of oil wells and grassland type

Baird’s Sparrow and Brown-headed Cowbird were less abundant in planted

pastures with oil wells compared to native pastures with oil wells present. Sprague’s

Pipit was less abundant in planted pastures with oil wells, particularly when well density

was <7 wells/259 ha. Chestnut-collared Longspur was also less abundant near oil wells

in planted pastures compared to native pastures with wells. This suggests that planted

pastures may represent poorer quality habitat that compounds the negative effects of oil

development for these species. Kalyn-Bogard and Davis (2014) found relatively weak

relationships between bird abundance and gas well proximity and density and speculated

this may be due to high quality habitat in their study sites. Baird’s Sparrow, Chestnut-

collared Longspur and Sprague’s Pipit are typically considered grassland specialists, and

grasslands dominated by exotic grasses typically represent poorer-quality habitat for

grassland specialists compared to native pastures (Lloyd and Martin 2005, Fisher and

Davis 2011) resulting in reduced bird abundance (Dale et al. 1997, Flanders et al. 2006).

Thus, negative effects of oil development are likely greater in planted grasslands for

some species due to poorer quality habitat. However, Brown-headed Cowbird is

34

typically thought of as a generalist species with no preference for native or planted

grasslands (Shaffer et al. 2003), so it is unclear why cowbird abundance was greatest in

native pastures with abandoned wells. Possibly, the vegetation structure in native

pastures with abandoned wells better enables cowbirds to locate nests to parasitize.

In contrast to cowbirds, Bobolink exhibited reduced abundance in native pastures

with oil wells compared to planted grasslands. Planted grasslands are typically

associated with higher visual obstruction and vegetation height, especially hayfields in

the spring, which Bobolinks are known to prefer (Winter et al. 2005). Thus, it is possible

that planted grasslands (particularly hayfields) in my study area could represent better

quality breeding habitat for this species. This may not be the case for Bobolinks across

their entire breeding range, especially if fields are mowed (Bollinger 1995, Herkert 1997).

2.4.3. Cumulative effects of oil development on grassland songbird abundance

Cumulative effects are the temporally and spatially additive effects of similar

disturbance features or the interactive effects of multiple different disturbance features

that alter the structure or function of the ecosystem (Spaling and Smit 1993). While

individual disturbance features may have little effect on habitat quality or bird abundance,

the cumulative effects may be quite large (Spaling and Smit 1993). The cumulative

disturbance caused by oil development (overall amount of well pads, roads, pipelines,

building and battery pads) influenced the abundance of four species of grassland

songbirds. Bobolink and Savannah Sparrow had reduced abundance in areas with more

cumulative disturbance. Gaudet (2013) found that Savannah Sparrow had lower nest

survival near disturbance features associated with natural gas development. However,

several studies have found that Savannah Sparrow abundance was greatest in areas with

35

increased natural gas disturbance (Dale et al. 2009, Hamilton et al. 2011, Kalyn-Bogard

2011). Thus, my findings suggest that while natural gas development may not reduce the

quantity and quality of habitat for Savannah Sparrows, oil development does.

Grasshopper Sparrow abundance was greatest when there was a moderate level of

cumulative disturbance, but Clay-colored Sparrow abundance was lowest when there was

a moderate level of cumulative disturbance. This points to the potential for a limiting

relationship between the total amount of disturbance and habitat features for these species,

whereby above a certain threshold of disturbance (~3%) certain habitat features become

either attractive or unattractive.

Increased cumulative disturbance was generally associated with higher well

density, suggesting that there is likely increased traffic and anthropogenic activity in

more disturbed areas. Birds may be responding to this type of increased activity

(Ingelfinger and Anderson 2004, Lawson et al. 2011) and may experience reduced

reproductive success in areas with more disturbance and greater anthropogenic activity

(Lyon and Anderson 2003). Predator occurrence and predation risk may also increase in

areas with greater disturbance due to anthropogenic activity (Bui et al. 2010), linear

features that act as corridors for predators (Winter et al. 2000, Chalfound et al. 2002), or

vertical structures that provide avian predators with perches (Knight and Kawashima

1993). Grassland birds may also avoid highly disturbed areas because they are area

sensitive (Herkert et al. 2003, Davis 2004, Davis et al. 2006) or because of edge effects

(Bollinger and Gavin 2004, Ingelfinger and Anderson 2004, Koper et al. 2009). However,

some species that responded negatively to cumulative disturbance (e.g., Savannah

Sparrow) are not known to be area sensitive or respond negatively to edges.

36

Abundance of three species was also influence by single landscape disturbance

features associated with oil development. Baird’s Sparrow avoided areas with increased

well pad disturbance, and Western Meadowlark exhibited reduced abundance as the

amount of oil roads increased. Other studies have found that grassland songbirds avoid

roads and trails associated with natural gas development (Ingelfinger and Anderson 2004,

Kalyn-Bogard 2011, Ludlow et al. 2015), and Kalyn-Bogard (2011) found that Baird’s

Sparrow was influenced by natural gas well pad area. There was also an interactive

effect on meadowlark abundance between habitat and oil roads, whereby abundance was

lowest in planted pastures with more oil roads but there was no effect in native pastures.

This suggests that planted grasslands disturbed by oil roads represent poorer quality

habitat for Western Meadowlark. In contrast, Grasshopper Sparrow abundance increased

as the amount of oil battery and building pads increased, even though abundance was

reduced beyond a certain threshold of cumulative disturbance. These relationships

between single disturbance features do not indicate that any one type of disturbance

associated with oil development has a more pronounced effect on bird abundance.

Instead, my results suggest that it is the cumulative amount of disturbance that has the

greatest effect on abundance for at least four species.

It is also important to note the scale at which these disturbance features influences

bird abundance. Total cumulative disturbance was never higher than ~8% of my 259 ha

sample sites, and several species exhibited lower abundances stemming from single

disturbance features at <2% disturbance of the total landscape. These are not large-scale

disturbance features yet they still have detectable effects on grassland songbird

abundance. While a small amount of disturbance may be tolerable for some species of

37

grassland songbirds, as the cumulative amount of disturbance increases the negative

effects on abundance quickly increase.

My results also suggest that oil development may have a stronger influence on

grassland songbirds than other types of energy development. Overall, the proximity,

activity, and density of oil wells negatively influenced the abundance of seven species,

including several grassland specialists, while two generalist species (Brown-headed

Cowbird and Vesper Sparrow) were positively influenced by the presence or proximity of

oil wells. Cumulative disturbance also had a generally negative effect on the abundance

of several species. Studies investigating the effects of natural gas development (Dale et

al. 2009, Hamilton et al. 2011, Gaudet 2013, Kalyn-Bogard and Davis 2014) and wind

energy (Devereaux et al. 2008, Bennett et al. 2014, Hale et al. 2014) on grassland

songbirds have found variable or weak effects. Oil development creates a different

footprint on the landscape compared to natural gas development. Oil development often

includes extra well sites for the injection and extraction of hydraulic fracturing solutions

and the extra trails associated with those sites. Pump-jacks move, an operator visits

active wells daily, and single-well batteries have heavy tanker truck traffic. Also, well

pad footprints are much larger for oil production compared to natural gas. Mine is one of

the first studies examining the specific effects of oil well proximity, density, and

cumulative disturbance on grassland songbird abundance. My results support the

hypothesis that oil development reduces habitat quality and quantity for grassland

specialists such as Baird’s Sparrow and Sprague’s Pipit. However, my results show that

oil development also negatively affects species that prefer planted pastures (i.e.

38

Bobolink) and generalist species such as Savannah Sparrow, even though other types of

energy development typically positively influence Savannah Sparrow.

2.5. CONCLUSIONS

It is evident from my results that oil wells and the disturbance associated with

them influence bird abundance. While the particular effects of oil development are

somewhat species-specific, there is a general overall trend that oil development

negatively influences abundance, particularly for grassland specialists. At least seven

species experienced reduced abundance closer to oil wells or in areas with high well-

density. Only two species, Brown-headed Cowbird and Vesper Sparrow, had increased

abundance in the presence of oil wells, while Clay-colored Sparrow and Horned Lark

abundance were not influenced by the presence of oil wells at all. No species exhibited

greater abundance in areas with higher well density. Furthermore, the amount of

cumulative disturbance associated with oil development within 259 ha influences

grassland songbird abundance. Three species had decreased abundance in areas with

higher cumulative disturbance, while two species exhibited reduced abundance when the

amount of land converted to well pads and oil roads increased. Grasshopper Sparrow

showed increased abundance as oil battery and building area increased, although, as

cumulative disturbance increased beyond a threshold of 3% of the landscape Grasshopper

Sparrow abundance was reduced.

There is also more evidence that the type of grassland habitat plays a role in how

some species respond to the effects of oil development. Two grassland specialists and

two generalist species exhibited reduced abundance in planted grasslands where oil

39

development was present compared to native pastures. However, Bobolink exhibited

increased abundance in planted grasslands with oil development compared to native

pastures. Thus, factors that determine good habitat quality in the presence of oil

development may not be the same for all species.

Oil development does influence vegetation structure, which could in turn affect

grassland songbird abundance. However, I accounted for vegetation effects in my

models suggesting there are other factors associated with oil development influencing

bird abundance that extend beyond changes to vegetation structure. What the specific

mechanisms are that influence abundance remains unclear. However, the negative effects

of oil development on grassland songbirds are likely a reflection of additional stresses on

an already stressed habitat. The fact that birds experience reduced abundance in the

presence of oil development (particularly in tame grassland for grassland specialists) is an

indication that habitat quality may be further reduced by the pressures of oil development.

Further studies are needed to understand the mechanisms behind how oil wells

and disturbance cause reduced abundance in grassland songbirds. Before and after

development studies are lacking in the literature and need to be conducted to compare

bird abundance, density, and reproductive success before and after oil development is

completed. More data on reproductive success of songbirds in relationship to well

proximity, density, and cumulative disturbance are also needed. While bird abundance

gives some information on habitat quality, it is not sufficient to determine if occupied

habitat around oil wells is an ecological trap (Bayne and Dale 2011). Birds can be

present and establish breeding territories around wells, but they may experience reduced

reproductive success as a result of oil development. There is also a need for studies that

40

pair observation-type data with experimental components. Field experiments need to be

designed and conducted to determine if noise, traffic, human activity, the presence of

vertical structures, or movement of pumpjacks play a role in how birds respond to oil

wells and disturbance.

2.6 MANAGEMENT IMPLICATIONS

Based on the results of my study, several recommendations can be made to land

managers faced with decisions regarding grassland songbird conservation in the midst of

oil development:

1) Efforts should be taken to limit well density and associated cumulative

disturbance in grassland habitat. Sprague’s Pipit and Western Meadowlark were

less abundant as well density increased, and no species were more abundant in

areas with increased well densities. Also, several songbird species were less

abundant closer to wells, and as well density increases the distance between wells

will decrease, so limiting well density will also benefit species that avoid areas

near wells.

2) Wells should be placed as close together as possible, preferably on the same pads.

This will serve to limit the effects of well proximity on grassland birds, as well as

limiting the amount of disturbance from oil roads and other disturbance features.

Localizing disturbance and limiting wells to smaller areas is possible with today’s

directional drilling technology.

3) Oil development should be avoided on native grasslands as much as possible.

Several species sensitive to oil development are more abundant on native

41

grasslands, likely because it represents better quality habitat. Reducing oil

development on native grasslands will preserve the functional quality of native

grasslands for these species.

4) Efforts need to be made to limit the impact to surrounding vegetation when oil

wells are drilled.

5) For a summary of species-specific effects of oil development and habitat type on

abundance see Table 2.11.

42

2.7. LITERATURE CITED

Aldridge, C. L., and M. S. Boyce. 2007. Linking occurrence and fitness to persistence:

Habitat-based approach for endangered Greater Sage-Grouse. Ecological

Applications 17:508–526.

Arnold, T. W. 2010. Uninformative parameters and model selection using Akaike’s

Information Criterion. The Journal of Wildlife Management 74:1175–1178.

Askins, R. A., F. Chávez-Ramírez, B. C. Dale, C. A. Haas, J. R. Herkert, F. L. Knopf,

and P. D. Vickery. 2007. Conservation of grassland birds in North America:

Understanding ecological processes in different regions: “Report of the AOU

committee on conservation”. Ornithological Monographs:iii–46.

Bayne, E. M., S. L. Van Wilgenburg, S. Boutin, and K. A. Hobson. 2005. Modeling and

field-testing of Ovenbird (Seiurus aurocapillus) responses to boreal forest dissection

by energy sector development at multiple spatial scales. Landscape Ecology 20:203–

216.

Bayne, E. M., and B. C. Dale. 2011. Effects of energy development on songbirds. Pages

95–114 in D. E. Naugle, editor. Energy Development and Wildlife Conservation

in Western North America. Island Press/Center for Resource Economics.

Bayne, E. M., L. Habib, and S. Boutin. 2008. Impacts of chronic anthropogenic noise

from energy-sector activity on abundance of songbirds in the Boreal Forest.

Conservation Biology 22:1186–1193.

Beckmann, J. P., K. Murray, R. G. Seidler and J. Berger. 2012. Human-mediated shifts in

animal habitat use: sequential changes in pronghorn use of a natural gas field in

Greater Yellowstone. Biological Conservation 147: 222-233.

43

Bennett, V. J., A. M. Hale, K. B. Karsten, C. E. Gordon, and B. J. Suson. 2014. Effect of

wind turbine proximity on nesting success in shrub-nesting birds. The American

Midland Naturalist 172:317–328.

Bollinger, E. K. 1995. Successional changes and habitat selection in hayfield bird

communities. The Auk 112:720–730.

Bollinger, E. K., and T. A. Gavin. 2004. Responses of nesting bobolinks (Dolichonyx

oryzivorus) to habitat edges. The Auk 121:767–776.

Brennan, L. A., and W. P. Kuvlesky. 2005. North American grassland birds: an unfolding

conservation crisis? The Journal of Wildlife Management 69:1–13.

Bui, T.-V. D., Marzluff, and B. Bedrosian. 2010. Common raven activity in relation to

land use in western Wyoming: implications for Greater Sage-grouse reproductive

success. The Condor 112:65–78.

Burnham, K. P., and D. R. Anderson. 1998. Model Selection and Inference: A Practical

Information-theoretic Approach. Springer. 353p.

Chalfoun, A. D., F. R. Thompson, and M. J. Ratnaswamy. 2002. Nest predators and

fragmentation: a review and meta-analysis. Conservation Biology 16:306–318.

Dale, B. C., P. A. Martin, and P. S. Taylor. 1997. Effects of hay management on

grassland songbirds in Saskatchewan. Wildlife Society Bulletin 25:616–626.

Dale, B. C., T. S. Wiens, and L. E. Hamilton. 2009. Abundance of three grassland

songbirds in an area of natural gas infill drilling in Alberta, Canada. Proceedings

of the Fourth International Partners in Flight Conference: Tundra to Tropics:194–

204.

44

Davis, S. K. 2004. Area sensitivity in grassland passerines: Effects of patch size,

patch shape, and vegetation structure on bird abundance and occurrence

in southern Saskatchewan. Auk 121:1130–1145.

Davis, S. K., R. M. Brigham, T. L. Shaffer, and P. C. James. 2006. Mixed-grass prairie

passerines exhibit weak and variable responses to patch size. Auk 123:807–821.

Davis, S. K., R. J. Fisher, S. L. Skinner, T. L. Shaffer, and R. M. Brigham. 2013.

Songbird abundance in native and planted grassland varies with type and amount of

grassland in the surrounding landscape. The Journal of Wildlife Management

77:908–919.

Devereux, C. L., M. J. H. Denny, and M. J. Whittingham. 2008. Minimal effects of wind

turbines on the distribution of wintering farmland birds. Journal of Applied Ecology

45:1689–1694.

Environment Canada. 2009. Petroleum industry activity guidelines for wildlife species at

risk in the Prairie and Northern region. Canadian Wildlife Service, Environment

Canada, Prairie and Northern Region, Edmonton Alberta. 64p.

Fisher, R. J., and S. K. Davis. 2011. Post-fledging dispersal, habitat use, and survival of

Sprague’s Pipits: Are planted grasslands a good substitute for native? Biological

Conservation. 144:263-271.

Fiske, I., and R. Chandler. 2011. unmarked: An R package for fitting hierarchical models

of wildlife occurrence and abundance. Journal of Statistical Software 43:1–23.

Flanders, A. A., W. P. Kuvlesky, D. C. Ruthven, R. E. Zaiglin, R. L. Bingham, T. E.

Fulbright, F. Hernández, and L. A. Brennan. 2006. Effects of invasive exotic

grasses on south Texas rangeland breeding birds. The Auk 123:171–182.

45

Francis, C. D., C. P. Ortega, and A. Cruz. 2009. Noise pollution changes avian

communities and species interactions. Current Biology 19:1415–1419.

Gaston, K. J., T. M. Blackburn, and K. K. Goldewijk. 2003. Habitat conversion and

global avian biodiversity loss. Proceedings of the Royal Society B-Biological

Sciences 270:1293–1300.

Gaudet, C. A. 2013. The effects of natural gas development on density, reproductive

success and nest survival of grassland songbirds in south-western Saskatchewan

(MSc. Thesis). University of Regina, SK., Canada.

George, A. D., T. J. O’Connell, K. R. Hickman, and D. M. Leslie. 2013. Food availability

in exotic grasslands: a potential mechanism for depauperate breeding assemblages.

The Wilson Journal of Ornithology 125:526–533.

Gilbert, M. M., and A. D. Chalfoun. 2011. Energy development affects populations of

sagebrush songbirds in Wyoming. Journal of Wildlife Management 75:816–824.

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.

Government of Saskatchewan. 2015. Oil and gas. Ministry of Economy and Resources.

Available at: http://www.economy.gov.sk.ca/oilandgas Accessed June 25, 2015.

Hale, A. M., E. S. Hatchett, J. A. Meyer, and V. J. Bennett. 2014. No evidence of

displacement due to wind turbines in breeding grassland songbirds. The Condor

116:472–482.

Hamilton, L. E., B. C. Dale, and C. A. Paszkowski. 2011. Effects of disturbance

associated with natural gas extraction on the occurrence of three grassland

songbirds. Avian Conservation and Ecology 6:7.

46

Harju, S. M., M. R. Dzialak, R. C. Taylor, L. D. Hayden-Wing, and J. B. Winstead. 2010.

Thresholds and time lags in effects of energy development on Greater Sage-grouse

populations. Journal of Wildlife Management 74:437–448.

Herkert, J. R. 1997. Bobolink Dolichonyx oryzivorus population decline in agricultureal

landscapes in the Midwestern USA. Biological Conservation 80:107-112.

Herkert, J. R., D. L. Reinking, D. A. Wiedenfeld, M. Winter, J. L. Zimmerman, W. E.

Jensen, E. J. Finck, R. R. Koford, D. H. Wolfe, S. K. Sherrod, and others. 2003.

Effects of prairie fragmentation on the nest success of breeding birds in the mid-

continental United States. Conservation Biology 17:587–594.

Holloran, M. J., B. C. Fedy, and J. Dahlke. 2015. Winter habitat use of Greater Sage-

grouse relative to activity levels at natural gas well pads. The Journal of Wildlife

Management 79:630–640.

Ingelfinger, F., and S. Anderson. 2004. Passerine response to roads associated with

natural gas extraction in a sagebrush steppe habitat. Western North American

Naturalist 64:385–395.

Johnson, D. H. 2008. In Defense of indices: The case of bird surveys. Journal of Wildlife

Management 72:857–868.

Johnson, D. H., and M. D. Schwartz. 1993. The Conservation Reserve Program: Habitat

for grassland birds. Great Plains Research 3:273–295.

Kalyn Bogard, H. J. 2011. Natural gas development and songbird abundance in

southwestern Saskatchewan: the impact of gas wells and cumulative disturbance

(MSc. Thesis). University of Regina, SK., Canada.

47

Kalyn Bogard, H. J., and S. K. Davis. 2014. Grassland songbirds exhibit variable

responses to the proximity and density of natural gas wells. The Journal of

Wildlife Management 78:471–482.

Knight, R. L., and J. Y. Kawashima. 1993. Responses of raven and red-tailed hawk

populations to linear right-of-ways. Journal of Wildlife Management 57:266–271.

Koper, N., K. Molloy, L. Leston, and J. Yoo. 2014. Effects of livestock grazing and well

construction on prairie vegetation structure surrounding shallow natural gas wells.

Environmental Management 54:1131–1138.

Koper, N., D. J. Walker, and J. Champagne. 2009. Nonlinear effects of distance to habitat

edge on Sprague’s Pipits in southern Alberta, Canada. Landscape Ecology

24:1287–1297.

Lawson, A. L., M. L. Morrison, and R. D. Slack. 2011. Impacts of oil and gas

development on wintering grassland birds at Padre Island National Seashore,

Texas. Southeastern Naturalist 10:303–320.

Leu, M., S. E. Hanser, and S. T. Knick. 2008. The human footprint in the west: A large-

scale analysis of anthropogenic impacts. Ecological Applications 18:1119–1139.

Lloyd, J. D., and T. E. Martin. 2005. Reproductive success of Chestnut-collared

Longspurs in native and exotic grassland. The Condor 107:363–374.

Ludlow, S. M., R. M. Brigham, and S. K. Davis. 2015. Oil and natural gas development

has mixed effects on the density and reproductive success of grassland songbirds.

The Condor 117:64–75.

Lyon, A. G., and S. H. Anderson. 2003. Potential gas development impacts on sage

grouse nest initiation and movement. Wildlife Society Bulletin 31:486–491.

48

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 J. A. Royle. 2005. Designing occupancy studies: General advice

and allocating survey effort. Journal of Applied Ecology 42:1105–1114.

McKercher, R. B., and B. Wolfe. 1986. Understanding western Canada's dominion land

survey system. Division of Extension and Community Relations report,

University of Saskatchewan, Saskatoon. 26 pp.

McMaster, D. G., and S. K. Davis. 2001. An evaluation of Canada’s Permanent Cover

Program: Habitat for grassland birds? Journal of Field Ornithology 72:195–210.

McMaster, D. G., J. H. Devries, and S. K. Davis. 2005. Grassland birds nesting in

haylands of southern Saskatchewan: Landscape influences and conservation

priorities. Journal of Wildlife Management 69:211–221.

Murphy, M. T., and F. Moore. 2003. Avian population trends within the evolving

agricultural landscape of eastern and central United States. The Auk 120:20–34.

Mineau, P., and M. Whiteside. 2013. Pesticide acute toxicity is a better correlate of U.S.

grassland bird declines than agricultural intensification. PLOS ONE 8:2.

Northrup, J. M., and G. Wittemyer. 2013. Characterising the impacts of emerging energy

development on wildlife, with an eye towards mitigation. Ecology Letters 16:112–

125.

Noss, R. F. 2000. High-risk ecosystems as foci for considering biodiversity and

ecological integrity in ecological risk assessments. Environmental Science & Policy

3:321–332.

49

Noss, R. F., E. T. LaRoe, and J. M. Scott. (1995). Endangered ecosystems of the United

States: a preliminary assessment of loss and degradation. Biological Report 28,

USDI National Biological Service, Washington, DC.

Ribic, C. A., R. R. Koford, J. R. Herkert, D. H. Johnson, N. D. Niemuth, D. E. Naugle, 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.

Robel, R. J., J. N. Briggs, A. D. Dayton, and L. C. Hulbert. 1970. Relationships between

visual obstruction measurements and weight of grassland vegetation. Journal of

Range Management 23:295–297.

Royle, J. A. 2004. N-mixture models for estimating population size from spatially

replicated counts. Biometrics 60:108–115.

Saskatchewan Research Council. 2015. FlySask.ca. Version 2.0. Saskatchewan

Geospatial Imagery Collaborative, http://www.flysask.ca, Accessed July 14, 2015.

Sauer, J. R., J. E. Hines, J. E. Fallon, K. L. Pardieck, D. J. Ziolkowski, Jr., and W. A.

Link. 2014. The North American breeding bird survey, results and analysis 1966 -

2013. Version 01.30.2015 USGS Patuxent Wildlife Research Center, Laurel, MD.

Sawyer, H., M. J. Kauffman, and R. M. Nielson. 2009. Influence of well pad activity on

winter habitat selection of mule deer. Journal of Wildlife Management 73:1052–

1061.

Shaffer, J. A., C. M. Goldade, M. F. Dinkins, D. H. Johnson, L. D. Igl, and B. R. Euliss.

2003. Brown-headed Cowbirds in grasslands: their habitats, hosts, and response to

management. USGS Northern Prairie Wildlife Research Center: 158.

50

Sliwinski, M. S., and N. Koper. 2012. Grassland bird responses to three edge types in a

fragmented mixed-grass prairie. Avian Conservation and Ecology 7:6.

Spaling, H., and B. Smit. 1993. Cumulative environmental change: conceptual

frameworks, evaluation approaches, and institutional perspectives. Environmental

Management 17:587–600.

Souther, S., Tingley, M. W., Popescu, V. D., Hayman, D. T. S., Ryan, M. E., Graves, T.

A., Hartl, B., and Terrell, K. 2014. Biotic impacts of energy development from

shale: research priorities and knowledge gaps. Frontiers in Ecology and the

Environment 12:330-338.

Sutter, G. C., and R. M. Brigham. 1998. Avifaunal and habitat changes resulting from

conversion of native prairie to crested wheat grass: patterns at songbird

community and species levels. Canadian Journal of Zoology 76:869–875.

Van Wilgenburg, S. L., K. A. Hobson, E. M. Bayne, and N. Koper. 2013. Estimated

avian nest loss associated with oil and gas exploration and extraction in the

western Canadian sedimentary basin. Avian Conservation and Ecology 8:9.

Webb, S. L., C. V. Olson, M. R. Dzialak, S. M. Harju, J. B. Winstead, and D. Lockman.

2012. Landscape features and weather influence nest survival of a ground-nesting

bird of conservation concern, the Greater Sage-grouse, in human-altered

environments. Ecological Processes 1:1–15.

Winter, M., D. H. Johnson, and J. Faaborg. 2000. Evidence for edge effects on multiple

levels in tallgrass prairie. Condor 102:256–266.

51

Winter, M., D. H. Johnson, and J. A. Shaffer. 2005. Variability in vegetation effects on

density and nesting success of grassland birds. Journal of Wildlife Management

69:185–197.

52

Table 2.1. Distribution of well densities in native and planted sample sites (n=243) in southeastern SK, in 2013-2014. ‘No Well’ =

zero wells, ‘Low Density’ = 1-4 wells, ‘Medium Density’ = 5-8 wells, ‘High Density’ = 9+ wells. ‘Active Well’ = total number of

sites with an actively producing focal well; ‘Abandoned Well’ = total number of sites with a focal well no longer producing oil.

Habitat No Well Low Density Medium Density

High Density

Total Active Well Abandoned Well

Native 28 32 29 23 (26)1 112 69 15 Planted 27 30 32 42 (48)1 131 82 22 TOTAL 55 62 61 65 243 1Maximum well density

53

Table 2.2. Mean species abundance ± SE (with percent occurrence) of eleven commonly detected grassland songbird species in native

and planted pastures in southeastern SK, in 2013 and 2014, and frequency of occurrence for both years and habitat types combined.

Species Habitat 2013 Mean ± SE 2014 Mean ± SE Frequency of occurrence1

Baird’s Sparrow native planted

1.3 ± 0.10 (70%) 0.5 ± 0.07 (35%)

0.5 ± 0.10 (31%) 0.2 ± 0.04 (17%)

0.38

Brown-head Cowbird native planted

2.3 ± 0.14 (87%) 2.3 ± 0.17 (89%)

3.8 ± 0.29 (89%) 3.2 ± 0.16 (94%)

0.90

Bobolink native planted

0.6 ± 0.07 (37%) 1.0 ± 0.12 (54%)

0.4 ± 0.11 (23%) 1.2 ± 0.12 (50%)

0.43

Chestnut-collared Longspur

native planted

0.6 ± 0.08 (35%) 0.02 ± 0.01 (2%)

0.3 ± 0.10 (16%) 0.01 ± 0.01 (1%)

0.14

Clay-colored Sparrow native planted

1.9 ± 0.12 (81%) 1.7 ± 0.12 (85%)

2.0 ± 0.13 (89%) 1.6 ± 0.09 (81%)

0.83

Grasshopper Sparrow native planted

0.8 ± 0.08 (50%) 0.6 ± 0.08 (40%)

0.4 ± 0.07 (27%) 0.5 ± 0.06 (35%)

0.39

Horned Lark native planted

0.4 ± 0.06 (29%) 0.1 ± 0.04 (11%)

0.3 ± 0.07 (27%) 0.2 ± 0.04 (17%)

0.21

Savannah Sparrow native planted

2.5 ± 0.09 (97%) 2.3 ± 0.13 (93%)

1.8 ± 0.10 (94%) 2.3 ± 0.08 (98%)

0.96

Sprague’s Pipit native planted

0.6 ± 0.05 (50%) 0.1 ± 0.03 (5%)

0.4 ± 0.06 (36%) 0.1 ± 0.02 (10%)

0.24

Vesper Sparrow native planted

0.5 ± 0.06 (37%) 0.7 ± 0.06 (57%)

0.6 ± 0.09 (46%) 0.7 ± 0.05 (57%)

0.49

Western Meadowlark native planted

1.1 ± 0.07 (80%) 0.7 ± 0.08 (53%)

1.2 ± 0.08 (81%) 0.8 ± 0.05 (65%)

0.70

1Frequency of occurrence is for both years and habitat types combined

54

Table 2.3. Mean values (± SE) for vegetation variables recorded in native and planted pastures in southeastern SK in 2013-2014,

(native=224, planted=262).

Variable Habitat Mean ± SE p-value Litter depth (mm) native

planted 8.2 ± 0.1 7.7 ± 0.1

0.504

Vegetation height (cm) native planted

20.0 ± 0.7 23.0 ± 0.7

*0.002

Visual obstruction (Robel) native planted

9.0 ± 0.4 10.9 ± 0.4

*0.001

Total native grass cover (%) native planted

62.7 ± 2.0 3.1 ± 0.6

*0.000

Total exotic grass cover (%) native planted

11.2 ± 1.6 60.7 ± 1.6

*0.000

Total forb cover (%) native planted

9.6 ± 0.7 13.3 ± 0.9

*0.001

Shrub distance (m) native planted

2.1 ± 0.8 25.4 ± 1.4

*0.000

Bare ground (%) native planted

15.7 ± 1.3 22.2 ± 1.4

*0.001

*Statistically significant at p<0.05.

55

Table 2.4. Vegetation structure models relating distance to oil well (DIST), oil well density (DENS), and habitat type (native vs.

planted; HABITAT) to vegetation parameters in southeastern SK, 2013-2014 (n=486 survey sites). Interactions between habitat type

and oil well variables include all main effects. Models for each parameter are shown with the log likelihood score (LL), number of

parameters (K), Akaike’s Information Criterion score adjusted for small sample sizes (AICc), ∆AICc, and model weight (wi).

Model LL K AICc ∆AICc wi Bare Ground DIST*HABITAT aDIST DIST+DENS DIST*HABITAT+DENS*HABITAT DENS DENS*HABITAT NULL

-178.9 -181.4 -181.0 -178.3 -183.2 -181.5 -185.5

4 2 3 6 2 4 1

365.8 366.8 368.0 368.7 370.5 371.1 373.1

0.0 1.0 2.2 2.9 4.7 5.2 7.3

0.42 0.26 0.14 0.10 0.04 0.03 0.01

Forbs aNULL DIST DENS DIST+DENS DIST*HABITAT DENS*HABITAT DIST*HABITAT+DENS*HABITAT

-72.0 -72.0 -72.1 -72.1 -71.4 -71.7 -71.6

1 2 2 3 4 4 6

145.9 148.0 148.3 150.3 150.9 151.5 155.4

0.0 2.0 2.4 4.4 5.0 5.6 9.5

0.52 0.19 0.16 0.06 0.04 0.03 0.00

Litter aNULL

-571.6

2

1147.1

0.0

0.38

56

DIST DENS DIST+DENS DIST*HABITAT DENS*HABITAT DIST*HABITAT+DENS*HABITAT

-570.9 -571.5 -570.9 -570.4 -570.7 -569.4

3 3 4 5 5 7

1147.8 1149.0 1149.9 1150.9 1151.6 1153.0

0.7 1.8 2.7 3.8 4.5 5.8

0.26 0.15 0.10 0.06 0.04 0.02

Native aDIST*HABITAT DIST*HABITAT+DENS*HABITAT DENS*HABITAT DIST+DENS DENS DIST NULL

-143.6 -142.4 -149.9 -304.9 -306.7 -307.5 -313.5

4 6 4 3 2 2 1

295.3 296.9 308.0 615.8 617.3 619.0 629.0

0.0 1.6

12.7 320.5 322.1 323.7 333.7

0.69 0.31 0.00 0.00 0.00 0.00 0.00

Robel DIST*HABITAT+DENS*HABITAT aDIST+DENS DENS*HABITAT DENS DIST*HABITAT NULL DIST

-1588.9 -1592.8 -1593.6 -1597.8 -1596.7 -1602.9 -1601.9

7 4 5 3 5 2 3

3191.9 3193.6 3197.4 3201.7 3203.5 3209.8 3209.9

0.0 1.7 5.4 9.8

11.5 17.9 18.0

0.660.290.040.000.000.000.00

Tame aDIST*HABITAT DIST*HABITAT+DENS*HABITAT DENS*HABITAT NULL

-236.2 -234.5 -238.8 -324.4

4 6 4 1

480.5 481.1 485.7 650.8

0.0 0.6 5.2

170.3

0.55 0.41 0.04 0.00

57

DENS DIST DIST+DENS

-324.2 -324.8 -324.1

2 2 3

652.3 653.6 654.2

171.9 173.1 173.7

0.00 0.00 0.00

Vegetation Height aDIST+DENS DIST*HABITAT+DENS*HABITAT DENS*HABITAT DENS DIST*HABITAT NULL DIST

-1821.5 -1818.5 -1822.8 -1825.8 -1827.0 -1832.3 -1831.9

4 7 5 3 5 2 3

3651.0 3651.2 3655.6 3657.7 3664.1 3668.7 3669.8

0.0 0.1 4.6 6.6

13.1 17.6 18.7

0.48 0.45 0.05 0.02 0.00 0.00 0.00

Shrub Distance aDIST*HABITAT DENS*HABITAT DIST*HABITAT+DENS*HABITAT DENS DIST+DENS DIST NULL

-2088.5 -2089.9 -2088.4 -2143.7 -2142.9 -2144.4 -2147.0

5 5 7 3 4 3 2

4187.2 4189.9 4191.0 4293.5 4293.9 4294.8 4298.0

0.0 2.8 3.8

106.3 106.7 107.6 110.8

0.71 0.18 0.10 0.00 0.00 0.00 0.00

aMost parsimonious model for each vegetation parameter; based on AICc score, wi, and number of parameters

58

Table 2.5. N-mixture abundance models comparing and combining the most parsimonious oil well models (well proximity [distance],

well density [density], well activity [activity]), habitat type [habitat], and vegetation model (native cover [native], exotic grass cover

[tame], litter depth [litter], forbs [forbs], shrub distance [shrubs], visual obstruction [robel], bare ground [bare], vegetation height

[height]), the observation only model [Detection] and the Null model in southeastern SK, 2013-2014. All oil, vegetation, and habitat

models include the observation model. All models are presented with the log likelihood score (LL), number of parameters (K),

Akaike’s Information Criterion score adjusted for small sample sizes (AICc), ∆AICc, and model weight (wi).

Model LL K AICc ∆AICC wi Baird’s Sparrow anative + litter + forbs2 + shrubs2 + distance + activity*habitat native + litter + forbs2 + shrubs2 + distance*habitat + activity*habitat native + litter + forbs2 + shrubs2 + distance*habitat + activity native + litter + forbs2 + shrubs2 + distance + activity + habitat native + litter + forbs2 + shrubs2 + distance + activity native + litter + forbs2 + shrubs2 distance + activity Detection Null

-1063.8 -1063.6 -1066.0 -1068.3 -1072.4 -1086.6 -1123.4 -1148.0 -1248.8

22 23 21 20 19 16 15 12 2

2173.9 2175.5 2176.0 2178.4 2184.3 2206.3 2277.8 2320.7 2501.6

0.0 1.6 2.1 4.6

10.5 32.4

103.9 146.8 327.8

0.53 0.23 0.18 0.05 0.00 0.00 0.00 0.00 0.00

Brown-headed Cowbird arobel2 + activity*habitat robel2 + activity + habitat robel2 + activity robel2 activity Detection

-3166.9 -3173.0 -3175.1 -3181.3 -3182.8 -3188.8

17 15 14 12 13 11

6369.0 6377.1 6379.1 6387.3 6392.4 6400.1

0.0 8.1

10.0 18.3 23.4 31.1

0.98 0.02 0.01 0.00 0.00 0.00

59

Null

-3283.6 2 6571.2 202.1 0.00

Bobolink atame + shrub2 + veg height + distance*habitat tame + shrub2 + veg height + distance + habitat tame + shrub2 + veg height + distance tame + shrub2 + veg height distance Detection Null

-1395.8 -1400.2 -1404.8 -1421.3 -1455.3 -1469.4 -1507.3

14 13 12 11 9 8 2

2820.5 2827.1 2834.2 2865.1 2929.0 2955.0 3018.5

0.0 6.6

13.7 44.6

108.5 134.5 198.1

0.96 0.04 0.00 0.00 0.00 0.00 0.00

Chestnut-collared Longspur atame + litter + forbs2 +shrubs2 + distance + habitat tame + litter + forbs2 + shrubs2 + distance*habitat tame + litter + forbs2 + shrubs2 tame + litter + forbs2 + shrubs2 + distance distance Detection Null

-415.2 -415.0 -426.0 -425.3 -492.1 -495.9 -572.9

17 18 15 16 12 11 2

865.8 867.5 883.0 883.8

1008.8 1014.4 1149.9

0.0 1.7

17.2 18.0

143.0 148.6 284.1

0.70 0.30 0.00 0.00 0.00 0.00 0.00

Clay-colored Sparrow ashrubs + robel + litter shrubs + robel + litter + habitat Detection Null

-2251.8 -2251.6 -2303.5 -2455.8

13 14 10 2

4530.5 4532.1 4627.4 4915.5

0.0 1.7

96.9 385.1

0.70 0.30 0.00 0.00

Grasshopper Sparrow (2013) arobel2 + distance2

distance2

robel2 + distance2 + habitat robel2 + distance2*habitat

-471.3 -473.6 -470.5 -468.9

10 9

12 14

963.5 966.0 966.4 967.6

0.0 2.5 2.9 4.1

0.60 0.18 0.14 0.08

60

robel2

Detection Null

-480.8 -483.1 -624.2

9 8 2

980.3 982.8

1252.5

16.8 19.3

289.0

0.00 0.00 0.00

Grasshopper Sparrow (2014) ashrubs + litter + activity + distance shrubs + litter + activity + distance + habitat shrubs + litter + activity*habitat + distance shrubs + litter + activity + distance*habitat shrubs + litter + activity*habitat + distance*habitat shrubs + litter activity + distance Detection Null

-382.2 -382.1 -382.1 -381.1 -381.1 -397.8 -398.1 -412.1 -482.1

13 14 15 16 17 10 11 8 2

792.0 794.1 796.3 796.6 798.9 816.5 819.4 840.8 968.3

0.0 2.1 4.3 4.6 6.9

24.5 27.4 48.7

176.3

0.62 0.22 0.07 0.06 0.02 0.00 0.00 0.00 0.00

Horned Lark atame + robel + litter + habitat tame + robel + litter Detection Null

-543.6 -547.7 -578.7 -609.3

13 12 9 2

1114.0 1120.1 1175.8 1222.6

0.0 6.2

61.8 108.6

0.96 0.04 0.00 0.00

Savannah Sparrow abare + robel2 + shrubs + distance bare + robel2 + shrubs + distance + habitat bare + robel2 + shrubs + distance*habitat bare + robel2 + shrubs distance Detection Null

-2609.9 -2609.8 -2609.0 -2612.4 -2622.2 -2626.2 -2699.2

16 17 18 15 13 12 2

5253.0 5254.9 5255.5 5255.9 5271.1 5277.1 5402.5

0.0 1.8 2.5 2.8

18.0 24.1

149.5

0.52 0.21 0.15 0.13 0.00 0.00 0.00

Sprague’s Pipit

61

anative + robel + litter + density2 + habitat native + robel + litter + density2*habitat native + robel + litter + density2 native + robel + litter density2 Detection Null

-528.2 -528.1 -537.3 -539.8 -583.1 -587.5 -634.0

15 16 14 13 11 10 2

1087.5 1089.3 1103.5 1106.3 1188.8 1195.5 1272.0

0.0 1.8

16.0 18.8

101.4 108.0 184.6

0.71 0.29 0.00 0.00 0.00 0.00 0.00

Vesper Sparrow anative2 + tame2 + robel + distance native2 + tame2 + robel + distance + habitat native2 + tame2 + robel + distance*habitat native2 + tame2 + robel distance Detection Null

-1098.3 -1097.3 -1097.2 -1100.9 1113.3

-1119.8 -1162.0

14 15 16 13 11 10 2

2225.5 2225.6 2227.5 2228.6 2248.1 2260.1 2328.0

0.0 0.1 2.0 3.1

23.6 34.6

102.5

0.39 0.39 0.14 0.08 0.00 0.00 0.00

Western Meadowlark anative + litter2 + robel + density2 + habitat native + litter2 + robel + density2*habitat native + litter2 + robel + density2

native + litter2 + robel density2 Detection Null

-1366.2 -1365.7 -1369.8 -1373.9 -1380.9 -1386.9 -1514.1

16 17 15 14 12 11 2

2765.5 2766.8 2770.6 2776.7 2786.4 2796.4 3032.3

0.0 1.3 5.2

11.2 20.9 30.9

266.8

0.62 0.33 0.05 0.00 0.00 0.00 0.00

aTop overall abundance model for each species

62

Table 2.6. Parameter estimates of the top N-mixture model for abundance of eleven species of grassland songbirds in southeastern SK,

2013-2014.

Parameter Estimate Standard Error

Lower 85% CI

Upper 85% CI

Baird’s Sparrow Intercept anative alitter aforbs2

shrubs2

adistance habitat (native) ahabitat (tame) activity (abandoned) aactivity (active) activity (no well) activity*habitat (abandoned*native) activity*habitat (abandoned*tame) activity*habitat (active*native) aactivity*habitat(active*tame) activity*habitat(no well*native) aactivity*habitat(no well*tame)

-0.0091 0.2383

-0.1618 -0.1261 -0.0996 0.4480

0 -2.1176

0 0.4049

-0.1274 0 0 0

1.4072 0

1.8272

0.2626 0.0928 0.0654 0.0682 0.0796 0.1221

0 0.7607

0 0.2462 0.3405

0 0 0

0.7634 0

0.7687

-0.3872 0.1047

-0.2560 -0.2243 -0.2142 0.2722

0 -3.2130

0 0.0504

-0.6177 0 0 0

0.3079 0

0.7203

0.3690 0.3719

-0.0676 -0.0279 0.0150 0.6238

0 -1.0222

0 0.7594 0.3629

0 0 0

2.5065 0

2.9341

Brown-headed Cowbird aIntercept arobel2 habitat (native) ahabitat (tame) activity (abandoned)

2.7364 0.0479

0 -0.3822

0

0.1312 0.0104

0 0.1032

0

2.5475 0.0329

0 -0.5308

0

2.9253 0.0629

0 -0.2336

0

63

activity (active) aactivity (no well) activity*habitat (abandoned*native) activity*habitat (abandoned*tame) activity*habitat (active*native) aactivity*habitat(active*tame) activity*habitat(no well*native) aactivity*habitat(no well*tame)

0.0479 -0.3654

0 0 0

0.3121 0

0.5071

0.0839 0.1356

0 0 0

0.1180 0

0.1475

-0.0729 -0.5607

0 0 0

0.1422 0

0.2947

0.1687 -0.1701

0 0 0

0.4820 0

0.7195

Bobolink aIntercept atame ashrubs2 aveg height habitat (native) ahabitat (tame) distance*habitat(native) adistance*habitat(tame)

-0.477 0.145 0.110 0.291

0 0.544

0 -0.292

0.1217 0.0673 0.0475 0.0464

0 0.1596

0 0.0984

-0.6522 0.0481 0.0416 0.2242

0 0.3142

0 -0.4337

-0.3018 0.2419 0.1784 0.3578

0 0.7738

0 -0.1503

Clay-colored Sparrow aIntercept ashrubs arobel alitter

0.9162

-0.3077 0.1386 0.0751

0.0524 0.0382 0.0317 0.0299

0.8407

-0.3627 0.0930 0.0320

0.9917

-0.2527 0.1842 0.1182

Grasshopper Sparrow (2013) aIntercept arobel2 adistance2

1.270

-0.113 -0.313

0.2498 0.0622 0.0781

0.9103

-0.2026 -0.4255

1.6297

-0.0234 -0.2005

Grasshopper Sparrow (2014)

64

aIntercept ashrubs alitter adistance activity (abandoned) aactivity (active) activity (no well)

-1.079 0.198

-0.662 0.755

0 0.697

-0.568

0.260 0.094 0.160 0.168

0 0.273 0.462

-1.4534 0.0626

-0.8924 0.5131

0 0.3039

-1.2333

-0.7046 0.3334

-0.4316 0.9969

0 1.0901 0.0973

Chestnut-collared Longspur aIntercept atame alitter aforbs2

shrubs2

adistance habitat (native ahabitat (tame)

-0.566 -0.584 -0.513 -0.362 -0.135 0.141

0 -2.306

0.2777 0.2324 0.1535 0.2089 0.1760 0.0934

0 0.6015

-0.9659 -0.9187 -0.7340 -0.6628 -0.3884 0.0065

0 -3.1722

-0.1661 -0.2493 -0.2920 -0.0612 0.1184 0.2755

0 -1.4398

Horned Lark aIntercept tame arobel alitter habitat (native) habitat (tame)

-0.6100 -0.0278 -0.4939 -0.5102

0 -0.7208

0.194 0.137 0.131 0.152

0 0.252

-0.8894 -0.2251 -0.6825 -0.7291

0 -1.0837

-0.3306 0.1695

-0.3053 -0.2913

0 -0.3579

Savannah Sparrow aIntercept abare arobel2 ashrubs

1.3635

-0.0970 -0.0546 0.0825

0.0637 0.0305 0.0182 0.0278

1.2718

-0.1409 09.808 0.0425

1.4552

-0.0531 -0.0284 0.1225

65

adistance

0.0680 0.0304 0.0242 0.1118

Sprague’s Pipit aIntercept anative arobel alitter adensity2

ahabitat (native) ahabitat (tame)

-0.542 0.256

-0.343 -0.339 -0.212

0 -1.383

0.198 0.127 0.130 0.123 0.137

0 0.322

-0.8271 0.0731

-0.5302 -0.5161 -0.4093

0 -1.8467

-0.2569 0.4389

-0.1558 -0.1619 -0.0147

0 -0.9193

Vesper Sparrow aIntercept anative2 atame2 arobel adistance

0.415

-0.197 -0.247 -0.195 -0.145

0.1361 0.0832 0.0822 0.0613 0.0646

0.2190

-0.3168 -0.3654 -0.2833 -0.2380

0.6110

-0.0772 -0.1286 -0.1067 -0.0520

Western Meadowlark aIntercept native alitter2 arobel adensity2

habitat (native) ahabitat (tame)

1.18532

-0.00657 -0.04945 -0.09313 -0.07179

0 -0.39628

0.1874 0.0685 0.0225 0.0529 0.0327

0 0.1426

0.9155

-0.1052 -0.0819 -0.1693 -0.1189

0 -0.6016

1.4552 0.0921

-0.0171 -0.0170 -0.0247

0 -0.1909

a85% confidence interval does not include zero

66

Table 2.7. Mean percent of sample site area covered by disturbance features associated with oil development, divided between well

density categories, in southeastern SK, 2013-2014. Cumulative Disturbance is the sum of all other disturbance features.

Disturbance Habitat No Well (± SE)

Low (± SE)

Medium (± SE)

High (± SE)

Oil Roads (%) native planted

0.1 ± 0.03 0.1 ± 0.03

0.3 ± 0.03 0.2 ± 0.04

0.6 ± 0.05 0.5 ± 0.05

1.0 ± 0.06 0.9 ± 0.07

Pipelines (%) native planted

0.1 ± 0.06 0.0 ± 0.04

0.1 ± 0.04 0.1 ± 0.04

0.3 ± 0.12 0.2 ± 0.07

0.4 ± 0.14 0.1 ± 0.03

Well Pads (%) native planted

0.0 ± 0.01 0.1 ± 0.05

0.2 ± 0.03 0.3 ± 0.08

0.5 ± 0.07 0.6 ± 0.09

0.8 ± 0.08 1.0 ± 0.10

Battery & Building Pads (%) native planted

0.0 ± 0.00 0.0 ± 0.02

0.1 ± 0.02 0.1 ± 0.05

0.2 ± 0.06 0.1 ± 0.04

0.3 ± 0.10 0.2 ± 0.04

Cumulative Disturbance (%) native planted

1.2 ± 0.27 2.3 ± 0.22

2.5 ± 0.31 2.8 ± 0.30

3.2 ± 0.35 3.9 ± 0.27

3.3 ± 0.31 4.8 ± 0.26

67

Table 2.8. Correlations between amount of area disturbed in sample sites (n=243) by oil development features in southeastern SK,

2013-2014.

Battery & Building

Pads

Oil Roads Pipelines Power lines

Cumulative Disturbance

Well Density

Well Pads

Battery & Building Pads

1.00 0.38 0.25 0.29 0.43 0.37 0.24

Oil Roads

0.38 1.00 0.19 0.49 0.39 0.77 0.48

Pipelines

0.25 0.19 1.00 0.21 0.28 0.06 0.09

Power lines

0.29 0.49 0.21 1.00 0.47 0.39 0.25

Cumulative Disturbance

0.43 0.39 0.28 0.47 1.00 0.57 0.54

Well Density

0.37 0.77 0.06 0.39 0.57 1.00 0.71

Well Pads 0.24 0.48 0.09 0.25 0.54 0.71 1.00

68

Table 2.9. N-mixture abundance models comparing and combining the most parsimonious landscape level disturbance models (area

disturbed by oil roads [oil roads], well pads [well pads], battery and oil building pads [bb pads], pipelines [pipelines], total sum of all

disturbance features [cumulative disturbance]), plus habitat type additive and interactive effects, and the observation only model

[Detection] and the Null model in southeastern SK, 2013-2014. All disturbance models include the observation model. All models

are presented with the log likelihood score (LL), number of parameters (K), Akaike’s Information Criterion score adjusted for small

sample sizes (AICc), ∆AICc, and model weight (wi).

Model LL K AICc ∆AICc wi Baird’s Sparrow awell pads + habitat well pads*habitat well pads Detection Null

-532.9 -532.8 -546.9 -550.9 -597.1

14 15 13 12

2

1095.5 1097.8 1121.4 1127.1 1198.2

0.0 2.2

25.8 31.6

102.7

0.75 0.25 0.00 0.00 0.00

Brown-headed Cowbird aDetection Null

-1600.2 -1645.2

11

2

3223.5 3294.5

0.0

71.0

1.00 0.00

Bobolink acumulative disturbance + habitat cumulative disturbance*habitat cumulative disturbance Detection Null

-606.0 -605.0 -622.1 -627.6 -657.0

10 11

9 8 2

1232.9 1233.2 1262.9 1271.8 1318.0

0.0 0.2

30.0 38.9 85.0

0.53 0.47 0.00 0.00 0.00

69

Chestnut-collared Longspur awell pads2 + habitat well pads2*habitat well pads2 Detection Null

-167.1 -167.1 -183.4 -187.0 -224.0

13 14 12 11

2

361.8 364.0 392.1 397.2 452.0

0.0 2.1

30.3 35.3 90.2

0.74 0.26 0.00 0.00 0.00

Clay-colored Sparrow acumulative disturbance2 + habitat cumulative disturbance2*habitat cumulative disturbance2

Detection Null

-1120.8 -1120.8 -1125.2 -1128.5 -1209.5

12 13 11 10

2

2267.0 2269.2 2273.5 2277.9 2423.0

0.0 2.2 6.6

10.9 156.0

0.73 0.24 0.03 0.00 0.00

Grasshopper Sparrow abb pads2 + cumulative disturbance2 bb pads2 + cumulative disturbance2 + habitat bb pads2 + cumulative disturbance2*habitat bb pads2*habitat + cumulative disturbance2 Detection Null

-367.8 -366.8 -366.5 -369.0 -373.6 -460.9

13 14 15 15 11

2

763.2 763.4 765.1 770.1 770.4 925.8

0.0 0.1 1.8 6.9 7.2

162.6

0.42 0.39 0.17 0.01 0.01 0.00

Horned Lark aDetection Null

-311.7 -320.4

9 2

642.2 644.8

0.0 2.6

0.79 0.21

Savannah Sparrow acumulative disturbance cumulative disturbance*habitat cumulative + habitat

-1278.0 -1276.3 -1278.0

13 15 14

2583.5 2584.8 2585.7

0.0 1.3 2.3

0.52 0.28 0.17

70

Detection Null

-1282.0 -1316.9

12 2

2589.3 2637.8

5.8 54.3

0.03 0.00

Sprague’s Pipit aDetection Null

-261.7 -285.4

10

2

544.3 574.8

0.0

30.5

1.00 0.00

Vesper Sparrow aDetection Null

-595.6 -628.4

10

2

1212.1 1260.8

0.0

48.7

1.00 0.00

Western Meadowlark aoil roads2*habitat oil roads + habitat oil roads Detection Null

-703.4 -704.7 -709.1 -711.6 -771.5

14 13 12 11

2

1436.7 1437.0 1443.6 1446.4 1547.1

0.0 0.3 6.9 9.7

110.4

0.53 0.45 0.02 0.00 0.00

aTop disturbance model for each species

71

Table 2.10. Parameter estimates of the top landscape disturbance N-mixture models of eleven species of grassland songbirds in

southeastern SK, 2013-2014.

Parameter Estimate Standard Error

Lower 85% CI

Upper 85% CI

Baird’s Sparrow aIntercept awell pads habitat (native) ahabitat (tame)

0.430

-0.260 0

-0.927

0.164 0.141

0 0.184

0.1938

-0.4630 0

-1.1920

0.6662

-0.0570 0

-0.6620

Brown-headed Cowbird aIntercept

2.52

0.174

2.2694

2.7706

Bobolink aIntercept acumulative disturbance habitat (native) ahabitat (tame)

-0.780 -0.405

0 0.949

0.1540 0.0854

0 0.1739

-1.0018 -0.5280

0 0.6986

-0.5582 -0.2820

0 1.1994

Chestnut-collared Longspur Intercept well pads2 habitat (native) ahabitat (tame)

-0.187 -0.408

0 -2.741

0.265 0.383

0 0.728

-0.5686 -0.9595

0 -3.7893

0.1946 0.1435

0 -1.6927

Clay-colored Sparrow aIntercept acumulative disturbance2 habitat (native)

0.9594 0.0905

0

0.0926 0.0332

0

0.8261 0.0427

0

1.0927 0.1383

0

72

ahabitat (tame) -0.2779

0.0938

-0.4130 -0.1428

Grasshopper Sparrow Intercept abuilding pads2 acumulative disturbance2

0.0983 0.0347

-0.2355

0.1840 0.0107 0.1021

-0.1667 0.0193

-0.3825

0.3633 0.0501

-0.0885

Horned Lark aIntercept

-0.639

0.173

-0.8881

-0.3899

Savannah Sparrow aIntercept acumulative disturbance

1.252

-0.129

0.0809 0.0455

1.1355

-0.1945

1.3685

-0.0635

Sprague’s Pipit aIntercept

-1.11

0.161

-1.3418

-0.8782

Vesper Sparrow Intercept

0.109

0.126

-0.0724

0.2904

Western Meadowlark aIntercept oil roads2 habitat (native) habitat (tame) oil roads2*habitat (native) aoil roads2*habitat (tame)

0.8388

-0.0216 0

-0.2204 0

-0.2185

0.1873 0.0855

0 0.1709

0 0.1403

0.5691

-0.1447 0

-0.4665 0

-0.4205

1.1085 0.1015

0 0.0257

0 -0.0165

a85% confidence interval does not include zero

73

Table 2.11. Summary table of effects of oil development features and habitat on grassland songbird abundance.

Species Well Distance Well Density Well Activity Landscape Disturbance

Habitat Type

Baird’s Sparrow reduced abundance near wells

NE* increased abundance at active wells compared to abandoned

reduced abundance with increased well pad area

planted grasslands reduces abundance in oil developed areas

Brown-headed Cowbird

NE NE increased abundance with abandoned wells

NE native grasslands increases abundance in oil developed areas

Bobolink reduced abundance near wells

NE NE reduced abundance with increased cumulative disturbance

planted grasslands increases abudnance in oil developed areas

Chestnut-collared Longpsur

reduced abundance near wells

NE NE NE planted grasslands reduces abundance in oil developed areas

Clay-colored Sparrow

NE NE NE least abundant at ~3% cumulative disturbance

NE

Grasshopper Sparrow

max abundance at ~450 m from wells

NE increased abundance when

max abundance at ~3% cumulative

NE

74

active well present disturbance; increased abundance as battery pads increase

Horned Lark

NE NE NE NE NE

Savannah Sparrow reduced abundance near wells

NE NE reduced abundance with increased cumulative disturbance

NE

Sprague’s Pipit NE reduced abundance with increased well density

NE NE planted grasslands reduces abundance in oil developed areas

Vesper Sparrow increased abundance near wells

NE NE NE NE

Western Meadowlark

NE reduced abudance with increased well density

NE reduced abundance with increased oil trails

planted grasslands reduces abundance in oil developed areas

*NE = no effect

75

Figure 2.1 Map of the study area, grass/pasture area, and 243 sample sites (with well densities) in southeastern SK, 2013-2014.

76

Figure 2.2 Typical sample site showing the 908 m, 400 m and 100 m buffers, along with

the locations of avian point-counts and oil wells.

77

Figure 2.3. Predicted change in proportion of bare ground cover (± 85% CI) in relationship to oil well proximity in southeastern SK,

2013-2014.

78

Figure 2.4. Predicted change in proportion of native cover (± 85% CI) in relationship to oil well proximity in southeastern SK, 2013-

2014.

79

Figure 2.5. Predicted change in proportion of exotic grass cover (± 85% CI) in relationship to oil well proximity in southeastern SK,

2013-2014.

80

Figure 2.6. Predicted change in visual obstruction and density (Robel; ± 85% CI) in relationship to oil well density and well proximity

in southeastern SK, 2013-2014.

Wellnumber Logwelldistance

81

Figure 2.7. Predicted change in vegetation height (± 85% CI) in relationship to oil well density and well proximity in southeastern SK,

2013-2014.

Wellnumber Logwelldistance

82

Figure 2.8. Predicted change in distance to the nearest shrub (± 85% CI) in relationship to oil well proximity in southeastern SK,

2013-2014.

83

Figure 2.9. Model-predicted mean abundance (± 85% CI) of Baird’s Sparrow varied with well proximity at active and abandoned well

sites in native and planted pastures in southeastern SK, 2013-2014.

84

Figure 2.10. Model-predicted mean abundance (± 85% CI) of Baird’s Sparrow varied with well activity between native and planted

pastures, and with litter depth, forb cover, and exotic grass cover in southeastern SK, 2013-2014.

85

Figure 2.11. Model-predicted mean abundance (± 85% CI) of Brown-headed Cowbird varied with well activity between native and

planted pastures, and with visual obstruction in southeastern SK, 2013-2014.

86

Figure 2.12. Model-predicted mean abundance (± 85% CI) of Bobolink varied with well distance between native and planted pastures,

and with vegetation height, shrub distance, and exotic grass cover in southeastern SK, 2013-2014.

87

Figure 2.13. Model-predicted mean abundance (± 85% CI) of Chestnut-collared Longspurs varied with well proximity, habitat type,

litter depth, forb cover, and exotic grass cover in southeastern SK, 2013-2014.

88

Figure 2.14. Model-predicted mean abundance (± 85% CI) of Grasshopper Sparrows in 2013 varied with well proximity and visual

obstruction in southeastern SK.

89

Figure 2.15. Model-predicted mean abundance (± 85% CI) of Grasshopper Sparrows in 2014 varied with well proximity, well activity,

and litter depth and shrub distance in southeastern SK.

90

Figure 2.16. Model-predicted mean abundance (± 85% CI) of Savannah Sparrow varied with well proximity, visual obstruction, shrub

distance, and bare ground cover in southeastern SK, 2013-2014.

91

Figure 2.17. Model-predicted mean abundance (± 85% CI) of Vesper Sparrow varied with well proximity, visual obstruction, and

native and exotic grass cover in southeastern SK, 2013-2014.

92

Figure 2.18. Model-predicted mean abundance (± 85% CI) of Sprague’s Pipit varied with well density, habitat type, litter depth,

native cover, and visual obstruction in southeastern SK, 2013-2014.

93

Figure 2.19. Model-predicted mean abundance (± 85% CI) of Western Meadowlark varied with well density in native and planted

pastures, litter depth, and visual obstruction in southeastern SK, 2013-2014.

94

Figure 2.20. Model-predicted mean abundance (± 85% CI) of Clay-colored Sparrow varied with shrub distance, visual obstruction

and litter depth in southeastern SK, 2013-2014.

95

Figure 2.21. Model-predicted mean abundance (± 85% CI) of Horned Lark varied with percent litter depth, visual obstruction, and

habitat type in southeastern SK, 2013-2014.

96

Figure 2.22. Correlation between well density and total area disturbed by oil development in samples (n=243) in southeastern SK,

2013-2014.

0 10 20 30 40

02

46

8

Well Density

Tota

l Dis

turb

ance

(%)

r = 0.57

97

Figure 2.23. Model-predicted mean abundance (± 85% CI) of Bobolink, Clay-colored Sparrow, Grasshopper Sparrow and Savannah

Sparrow at the landscape scale varied with cumulative disturbance in southeastern SK, 2013-2014.

98

Figure 2.24. Model-predicted mean abundance (± 85% CI) of Baird’s Sparrow at the landscape scale varied with percent area

disturbed by well pads in native and planted pastures in southeastern SK, 2013-2014.

99

Figure 2.25. Model-predicted mean abundance (± 85% CI) of Grasshopper Sparrow at the landscape scale varied with percent area

disturbed by battery and oil building pads in southeastern SK, 2013-2014.

100

Figure 2.26. Model-predicted mean abundance (± 85% CI) of Western Meadowlark at the landscape scale varied with percent area

disturbed by oil roads in native and planted pastures in southeastern SK, 2013-2014.

101

3. EFFECTS OF OIL WELL DENSITY AND OVERALL DISTURBANCE ON

GRASSLAND AVIAN PREDATOR OCCURRENCE

3.1 INTRODUCTION

As Europeans settled across the Great Plains they altered the landscape through

agriculture, urbanization, and resource development and extraction. Such anthropogenic

changes have provided many predators with increased opportunities for food, shelter and

water (Prugh et al. 2009). Consequently, many predators have increased in number and

expanded their range to include many parts of the Great Plains where they were

historically uncommon or absent (Boarman 1993, Sargeant et al. 1993, Prugh et al. 2009).

Furthermore, anthropogenic changes to the landscape can also influence the abundance,

activity and behaviour of grassland predators (Knight and Kawashima 1993, Phillips et al.

2003, Howe et al. 2014). Anthropogenic changes on the landscape often increases travel

corridors, fragmentation and habitat edges, which can attract predators and increase their

foraging efficiency (Johnson and Temple 1990, Dijak and Thompson 2000, Chalfoun et

al. 2002). In particular, predators of grassland birds have benefited from anthropogenic

subsidies, such as the Common Raven (Corvus corax; Boarman 1993, Boarman et al.

2006, Bui et al. 2010), and Striped Skunk (Mephitis mephitis; Lariviere et al. 1999).

Grassland songbird reproductive success is largely driven by predation on eggs

and chicks (henceforth nest predation; Martin 1988, Ribic et al. 2012), with grassland

nesting birds experiencing higher predation rates than above-ground nesting species

(Ricklefs 1969, Martin 1993). Studies employing nest cameras have shown that a wide

range of predators prey on grassland bird nests, including mid-sized mammals (e.g.

coyote, red fox, badger, skunk, raccoon), small mammals (e.g. ground squirrels, mice,

102

weasels, voles), deer, snakes, and birds (corvids, hawks) (Pietz and Granfors 2000,

Renfrew and Ribic 2003, Davis et al. 2012, Ribic et al. 2012).

Energy development can drastically alter landscapes by reducing habitat,

degrading habitat quality, and increasing fragmentation, edges and linear features

(Souther et al. 2014). Oil extraction is a common activity in North American grasslands

and this activity is rapidly expanding. Saskatchewan alone had an estimated 30,756

actively producing oil wells in 2014 (Government of Saskatchewan 2015), with

thousands of new wells being drilled every year. The potential impacts of oil

development on grassland ecosystems include linear (roads, pipelines, power lines, and

fences) and non-linear (wells, buildings, soil compaction, noise, traffic, and invasive

species) disturbances. Disturbance features associated with energy development in

grasslands may increase the occurrence and activity of grassland songbird predators

(Ingelfinger and Anderson 2004, Kalyn-Bogard and Davis 2014). For example, oil and

natural gas development has facilitated increased raven populations (Bui et al. 2010),

which reduces Greater Sage-grouse (Centrocercus urophasianus) nest success (Coates

and Delahanty 2004, 2010), particularly nests that are situated closer to oil wells and

other disturbance features (Webb et al. 2012, Dinkins 2013). Hawks and corvids may use

oil wells and other vertical structures to perch or nest on (Hall et al. 1981, Knight and

Kawashima 1993, Steenhof et al. 1993), making it easier to locate prey (Bui et al. 2010,

Keough and Conover 2012). Vegetation structure influences small mammal activity and

abundance (Dion et al. 2000, Thompson and Gese 2013), so altered vegetation structure

near wells and other disturbance features may lead to increased activity. Altered

vegetation, particularly short vegetation with increased bare ground, near oil disturbance

103

features may also increase predation rates on nests since avian and mammalian predators

have greater success when they can detect nests from a distance (Angelstam 1986). Mid-

sized mammalian predators (e.g. coyote, fox, raccoon, skunk, badger) may use linear

disturbance features (e.g. roads, pipelines) as travel corridors (Heske et al. 1999, Winter

et al. 2000, Renfrew and Ribic 2003, Renfrew et al. 2005), which could increase their

occurrence in areas with higher well density and increased cumulative disturbance.

Since predators are the main cause of reproductive failure and reduced survival in

grassland songbirds, changes to grassland predator activity due to oil development may

be influencing the declines of prairie bird populations. However, there is little published

research on the behaviour and activity of grassland songbird predators in response to oil

development. Thus, an understanding of the extent to which predators are influenced by

oil development will allow researchers and land managers to identify possible

mechanisms affecting grassland songbird demography in areas where oil extraction

occurs. The purpose of my research was to determine whether well density and

disturbance features associated with oil development influence the occurrence of

potential grassland songbird predators. I hypothesized that oil development provides

suitable foraging habitat for avian predators through increased number of perch sites.

Thus, I predicted that avian predator occurrence would be higher in areas with increased

well density and associated oil disturbance features.

3.2 METHODS

3.2.1 Study area

104

I conducted my study in southeastern Saskatchewan (Fig. 3.1) in 2013 and 2014,

in an area of active oil development and extraction along the edge of the moist-mixed

grasslands and aspen parkland ecoregions. The region is characterized by rolling

grassland dotted with aspen bluffs and wetlands. Most of the region has been cultivated,

producing a wide variety of cereals, oil seeds, feed grains, and forage crops. Native

vegetation is primarily confined to non-arable pasture lands, often bordering rivers and

streams. Native vegetation consists primarily of aspen (Populus tremuloides) bluffs,

shrubs such as western snowberry (Symphoricarpos occidentalis), prairie rose (Rosa

arkansana), chokecherry (Prunus virginiana), and wolf willow (Elaeagnus commutata),

and a mix of speargrasses (Aristida spp.) and wheatgrasses (Agropyron spp.). A large

portion of the study area has been seeded to exotic grasses and legumes for pasture and

hay. Planted pastures are typically composed of crested wheatgrass (Agropyron

cristatum), smooth brome (Bromus inermis), and Kentucky bluegrass (Poa pratensis) and

hayfields are typically characterized by smooth brome grass and alfalfa (Medicago spp.).

The area has experienced increasing oil development since 2008, when hydraulic

fracturing was introduced (Government of Saskatchewan 2015). As a result, oil wells in

the study area are a mix of conventional extraction and hydraulic fracturing.

3.2.2 Study site selection

I randomly selected oil wells across a gradient of oil-well densities in native and

tame pastures using a Geographic Information System (ArcGIS 10.0-10.2, ESRI). I

buffered each well by a 908 m radius (259 ha) sample site and quantified the density of

oil wells in each buffer. A well density gradient was classified based on four well-

density categories: none, low (1-4 wells/259 ha), medium (5-8/259 ha), and high (≥9/259

105

ha). These categories were chosen based on well-density categories used in similar

studies (Dale et al. 2009; Hamilton et al. 2011). I refer to the buffered areas as my

sample sites and the well at the center of each buffer as the focal well for each sample site.

Well density was calculated as the number of wells/259 ha because this represents the

size of a section of land (surveyed as 1 mile [1.61 km] x 1 mile; McKercher and Wolfe

1986), about which management decisions are often made. I ensured that sample sites

did not overlap.

Oil wells were divided into ‘active’ and ‘abandoned’ categories. An active well

consisted of a productive well with a pump-jack and gravel well pad. The size of well

pads ranged from traditional tear-drop shaped gravel pads to pads that had been stripped

of all top soil and piled in burms around the entire lease site. Also, some wells had oil

storage tanks located on the pad and some sites included more than one well on the same

pad. Pump-jacks at active wells were driven by relatively quiet electric motors. An

operator visited active well sites on a daily basis. Abandoned wells were non-productive

wells that ranged from a capped pipe with vegetation growing up to the pipe to sites with

minimal gravel pads and old pump-jack structures left in place.

Bird survey locations were randomly located within a 100 m buffer of each focal

well. Grassland habitat associated with each well location was verified through ground-

truthing following bird surveys. Habitat type (‘native’ vs. ‘planted’) was determined by

the dominant vegetation type of the quarter section (McKercher and Wolfe 1986) that

each bird survey was located in.

3.2.3 Avian surveys

106

I quantified bird occurrence using five-minute point-count surveys. Each point-

count location was surveyed three times using four different observers within two days of

each other to ensure that I was sampling a closed population (i.e. birds were not

emigrating or immigrating from the point-count location within the sampling period).

Repeated visits by different observers allowed me to calculate detection probabilities for

each species and therefore better estimate occupancy (MacKenzie et al. 2002, MacKenzie

and Royle 2005, Johnson 2008). Bird surveys were conducted during the breeding

season after breeding territories had been established (May 20 to July 7, 2013 and 2014).

No surveys were conducted after July 7 to ensure that only local breeders for that year

were included in the analyses. Survey locations were rotated every second day between

the western, middle and eastern portions of the study area. All point counts were

conducted between sunrise and 1400 hrs, when winds were <20 km/h and there was <1.0

mm of precipitation. Observers recorded all species of birds seen or heard inside the

habitat patch (quarter section) of the point-count location. I assumed that corvid and

hawk species seen inside the surveyed habitat patch were actively using that habitat.

Special notes were taken if any birds were seen perching or nesting on oil development

structures.

3.2.4 Disturbance measurements

I located all accessible roadways, pipelines, well sites, power lines, fences, oil

batteries, and other buildings and disturbance features associated with oil development

with a hand-held Global Positioning System (GPS) unit and converted the locations to

point, line and polygon features in ArcMap (ArcGIS 10.0-10.2, ESRI). I also used colour

digital ortho-photos taken from 2008-2012 at a 0.6 m resolution and a 1:1000 scale,

107

acquired from the Saskatchewan Geospatial Imagery Collaborative (Saskatchewan

Research Council 2015), to quantify and map disturbance features that were not

accessible on the ground. Linear features (roads and trails, pipelines) were buffered

based on the average width of 36 randomly selected oil roads/trails, and pipelines

respectively. The areas of all well pads, battery and building pads, oil roads and trails,

and pipelines were summed for each sample site along with the length of power lines. I

created a cumulative disturbance category by summing the areas of all well pads, battery

and building pads, oil roads and trails, and pipelines for each sample site.

3.2.5 Statistical analyses

I performed all analyses in R (v. 3.1.1. “Sock it to Me”, The R Foundation for

Statistical Computing, 2014).

I used the ‘occu’ function in the “unmarked” package (Fiske and Chandler 2011)

in R for all bird occurrence analyses. The ‘occu’ function uses the hierarchical models

developed by Royle (2004) to estimate occurrence from spatially replicated count data.

Hierarchical models have two ‘sides’: an observation model, which uses a Bernoulli

distribution to estimate a detection probability, and an occurrence model, which is

directly informed by the observation model and uses a Bernoulli distribution to estimate

the site-specific occupancy probability.

To select the top occurrence model for each species I compared the AICc values

of 1) a model composed of the top disturbance features associated with oil development,

2) a model composed of the top oil development disturbance features with the additive

and interactive effects of habitat type (native vs. planted), 3) a detection-only model, and

4) a null model. I first fit the best observation model for each bird species using my

108

occurrence data and four detection parameters (ordinal date of survey, time of survey,

wind speed during survey, and observer). I also included two site-specific parameters in

the observation models (well density, well distance) in case the presence, movement, or

noise from oil wells influenced bird detection. I used all corvid and hawk species that

were detected during at least 10% of point counts. I compared the interactive and

additive effects of year for each detection parameter to determine if I could combine the

two years of data. I then created a global model by comparing each detection parameter

to the Null model and retained the detection parameters that performed better than the

Null model. I fit all subset models of the global model and selected the top observation

model to be used as the detection component of the occurrence model for each species

(Appendix E). I ranked each model using AIC values and weights (Burnham and

Anderson 1998). I used 85% confidence limits to identify uninformative parameters

(Arnold 2010) and to determine the strength of effect for each informative parameter. I

considered the top fitted model to be the one with the highest AIC weight (wi). If there

were competing models (∆AICc < 2 and the same or fewer parameters as the model with

the highest AIC weight), I selected the most parsimonious model as the top model. I used

this method for model selection in all further analyses.

Once the best observation model was selected, I fit the best occurrence model for

each species using landscape disturbance features. I assessed well density and six

landscape-scale disturbance features associated with oil development to determine their

effect on avian predator occurrence. The landscape-scale features included the area of:

pipelines, battery and oil building pads, well pads, oil roads/trails, and cumulative

disturbance (sum of all disturbance features). The length of power lines was also

109

included as a landscape-scale disturbance feature. I first assessed correlations between

all disturbance features to determine if any disturbance types were highly correlated with

one another. I also evaluated the interactive and additive effects of year with each

disturbance parameter to determine if I could combine the two years of data. I compared

the effect of each disturbance parameter on bird occurrence to the detection model and

selected the disturbance variables that performed better than the detection model for each

species. I fitted all subsets of the remaining disturbance parameters and selected the top

disturbance model for each species. Any uninformative parameters (85% CI which

includes zero) included in the top occurrence models are not discussed.

3.3 RESULTS

3.3.1 General results

I conducted 243 point counts at 243 sample sites (112 native and 131 planted)

across southeastern Saskatchewan in 2013-2014 (Table 3.1). These were distributed

across a gradient of no wells (55), low density (62), medium density (61) and high

density (65) sample sites (Table 3.1).

Three corvid species (American Crow [Corvus brachyrhynchos], Black-billed

Magpie [Pica hudsonia], Common Raven) and three hawk species (Northern Harrier

[Circus cyaneus], Red-tailed Hawk [Buteo jamaicensis], Swainson’s Hawk [Buteo

swainsoni]) were recorded at >10% of the point counts. American Crow had the highest

frequency of occurrence (28%), while Common Raven had the lowest frequency of

occurrence (11%). All other predator species had occurrence frequencies of 12-15%.

110

All types of oil disturbance features increased as well density increased (Table

3.2), and subsequently cumulative disturbance also increased with well density (Fig. 3.3).

There was no difference between the amount of disturbance in native and planted sites

(Table 3.2). There was a positive correlation (r>0.70) between well density and well

pads and oil roads (Table 3.3); however, no other disturbance features showed any strong

correlations (r<0.54; Table 3.3).

3.3.2 Effects of well density and cumulative disturbance on avian predator

occurrence

Northern Harrier was the only species that had oil development variables included

in its top occurrence model; all other corvid and hawks species did not have disturbance

variables associated with their top occurrence models. The top occurrence model for

Northern Harrier included well density (Table 3.4) with the occurrence of the bird

decreasing with increasing well density (Fig. 3.4). American Crow occurrence in native

and planted grasslands was dependent on the year of the survey (Table 3.4) with

occurrence of crows higher in planted grasslands in 2013 compared to native pastures but

no difference in occurrence between habitat types in 2014 (Fig. 3.5). The observation

model best explained the variation in occurrence for Common Raven and Red-tailed

Hawk (Table 3.4). Detection of ravens and Red-tailed Hawk was dependent on the

observer and Red-tailed Hawk detection also increased with time of day (Appendix E).

The occurrence of Black-billed Magpie and Swainson’s Hawk was not related to any

detection, oil disturbance, or habitat variables (Table 3.4).

3.4 DISCUSSION

111

Northern Harrier was the only species affected by oil development in my study

area. Northern Harrier occurrence decreased in areas with higher well densities. Counter

to expectations, the occurrence of all other corvids and hawks in my study were not

influenced by oil wells or associated disturbance features. Thus, my results suggest that

oil development in grassland habitat does not play an important role in influencing the

occurrence of these avian predators.

Northern Harrier populations have experienced a steady decline of 1.0% per year

from 1966-2012 (Sauer et al. 2014). Rates of decline are steeper in Canada (2.3%) than

in the United States (0.3%; Saurer et al. 2014). My results suggest that oil development

in grassland habitat should be considered a potential factor contributing to the reduction

in Northern Harrier populations. Northern Harriers may be more affected by oil

development compared to other avian predators because they are ground-nesters

(Dechant et al. 2002) and because they hunt by flying low over the ground rather than

foraging from perches.

Oil development may reduce the amount of suitable habitat for Northern Harriers.

Northern Harriers are regarded as area-sensitive (Johnson 2001, Dechant et al. 2002), and

they are typically found in larger grassland patches (Walk and Warner 1999, Johnson and

Igl 2001) and in patches surrounded by landscapes with large amounts of grassland

habitat (Herkert et al. 1999). Higher well density may contribute to habitat fragmentation

and a reduction in patch size, which in turn could lead to grassland patches that are too

small to support Northern Harriers. Northern Harriers may also avoid areas with higher

well densities because of increased anthropogenic activity and traffic (Benitez-Lopez et

112

al. 2010), or because these areas support reduced prey abundance or possibly alter prey

composition.

Researchers have speculated that energy development might influence the

behaviour and activity of grassland predators, which in turn may contribute to decreased

abundance and reproductive success of grassland songbirds near wells and associated

disturbance features (Ingelfinger and Anderson 2004, Kalyn Bogard and Davis 2014).

There is no published research on the effects of oil development on grassland predator

behaviour and activity, however Ludlow et al. (2015) were the first to investigate the

influence of oil and natural gas development on grassland songbird reproductive success.

They found that access trails to well pads had negative consequences for reproductive

success for primary endemic species such as Sprague’s Pipit and Baird’s Sparrow.

Furthermore, several studies on Greater Sage-grouse have found increased nest predation

by hawks and ravens near natural gas wells and associated development features (Webb

et al. 2012), as well as increased predation when ravens and hawks are more abundant

(Coates and Delehanty 2010, Dinkins 2013). It may be that increased oil development

leads to increased nest and perch opportunities for hawks and corvids through the

creation of power poles, transmission lines, buildings, pump-jacks, and other vertical

structures. Knight and Kawashima (1993) found that Common Ravens were most

common along roads and power lines, and that raven and Red-tailed Hawk nests were

more abundant along power lines. They further speculated that power lines and power

poles offer superior perch and nest sites for Red-tailed Hawks (Knight and Kawashima

1993). Other studies have also found that ravens and hawks commonly nest on electrical

transmission line towers (Steenhof et al. 1993), and that hawks readily use artificial

113

perches (Hall et al. 1981). However, I did not find any associations between corvid and

hawk occurrence and power lines or oil roads and trails, nor did I observe any nests on

power poles, and my results do not provide evidence for increased avian predator activity

or abundance in areas with oil development. In fact, some hawks such as Northern

Harrier may be negatively affected by the type and amount of oil development present. It

should also be noted that nest camera studies have shown the majority of grassland

songbird nest predation is attributed to small mammals (Pietz and Granfors 2000, Davis

et al. 2012), so it is possible that the activity of other grassland predators (e.g. ground

squirrels, mice) could be more influenced by oil development than avian predators.

During the course of my study, I recorded only three observations of hawks

perching on power poles and I found one Great Horned Owl (Bubo virginianus) nest on

the top of a pump-jack oil well. There were no other observations of avian predators

nesting or perching on any type of oil disturbance feature. It may be that these types of

disturbance features do not attract corvids and hawks or that they do not gain any

advantage by using them. It is also possible that perch and nest sites are not a limiting

resource in my study area. My study area was located along the edge of the moist-mixed

grasslands and aspen parkland ecoregions, where aspen bluffs are common. These

wooded areas likely offer many locations for nests and perches. My study area also had

numerous farmsteads, features of which may provide suitable perch and nest sites for

corvids and hawks. Farmsteads may also be attractive to corvids as a source of food,

since these birds are often attracted to sites with anthropogenic garbage (Webb et al. 2004,

Boarman et al. 2006). Suitable perch and nests sites are likely limited in more arid

regions with fewer trees and are used more by avian predators. If true, I would expect a

114

noticeable effect of power lines and other vertical infrastructure associated with oil

development on the behaviour and abundance of avian predators.

While most grassland avian predators in my study did not respond to the

disturbance variables I measured, they may be responding to other landscape features that

I did not measure. For example, Red-tailed Hawk, Swainson’s Hawk and corvid

occurrence and activity may be more strongly influenced by the amount of woody

vegetation because this provides better nesting habitat and perch sites (Bednarz and

Dinsmore 1982, Gilmer and Stewart 1984). They may also be influenced by the amount

of wetland area or agricultural land (Schmutz 1987, Bechard et al. 1990, Preston 1990).

Home range sizes of hawk species in my study may not correspond to the scale at which I

investigated the effects of oil disturbance on their occurrence. Toland (1985) found male

Northern Harrier home ranges averaged 256 ha in Missouri, while Martin (1987) found

home ranges averaged 1,570 ha for males and 113 for females in Idaho. Babcock (1995)

found that the home range of Swainson’s Hawk averaged 4038.4 ha in California and

individuals foraged up to 22.5 km from the nest. It may be that the scale of disturbance

features I considered, or the scale at which I recorded occurrences was too small for the

hawk species in my study. Hawks may also be more strongly influenced by prey

availability and abundance (Luttich et al. 1970, Bechard 1982, Janes 1984). It is also

possible that corvids may be attracted to roads with higher traffic volume than the roads I

measured because more traffic could provide more carrion for these species to scavenge

(Knight and Kawashima 1993). However, given my results, it is also possible that oil

development has very little impact on the activity of most hawks and corvids.

115

3.5 CONCLUSIONS

My results provide evidence that Northern Harrier occurrence is negatively

influenced by oil development. Northern Harrier experienced reduced occurrence as well

density increased. However, the activity of other grassland avian predators in my study

were not affected by oil development. My results also show that grassland habitat type

(native and planted) did not influence the occurrence of most avian predators. American

Crow was the only species influenced by habitat type and occurrence was greater in

planted grasslands in one of the two years. These results suggest that oil development

and grassland habitat type are not important factors in determining the occurrence of

buteos and most corvid species in my study. It is possible that buteos and corvids

respond to other landscape features that I did not measure, such as the amount of woody

vegetation, wetlands and cropland. It is also possible that perch and nest sites are not a

limiting factor in my study area. It would be beneficial to investigate corvid and hawk

activity in more arid regions with oil development where perch sites are more limiting.

Concurrently, it would also be important to investigate whether the activity of predators

in regions of oil development actually influences the reproductive success of grassland

songbirds.

3.6 MANAGEMENT IMPLICATIONS

Northern Harrier occurrence not only declined with increased well density, but

they were affected by very low densities (e.g. 1-4 wells/259 ha), therefore it is important

to limit the number of wells for conserving Northern Harrier. Furthermore, Northern

116

Harriers are known to be area-sensitive, so limiting the amount of disturbance associated

with oil wells to reduce habitat fragmentation is likely important for this species.

Further studies are needed to investigate the effects of oil development on

grassland avian predator activity and behaviour to inform management decisions and

conservation policies. Tracking hawks and corvids using radio- or satellite-transmitters

would yield insights into which landscape and development features are associated with

nesting and foraging. In particular, tracking Northern Harriers may produce further

insights into what habitat features are attractive for this species in a landscape dominated

by oil development. Tracking would also be helpful to discover which perch sites are

most attractive for hawks and corvids in a landscape with an abundance of sites (e.g.

power poles, industrial features, wooded areas, farmyards). Finally, tracking the

movements of predators in a landscape with oil development would provide insight into

what impact they have on grassland songbird reproductive success.

117

3.7. LITERATURE CITED

Angelstam, P. 1986. Predation on ground-nesting birds’ nests in relation to predator

densities and habitat edge. Oikos 47:365–373.

Arnold, T. W. 2010. Uninformative parameters and model selection using Akaike’s

Information Criterion. The Journal of Wildlife Management 74:1175–1178.

Babcock, K. W. 1995. Home range and habitat use of breeding Swainson’s Hawks in the

Sacramento Valley of California. Journal of Raptor Research 29:193-197.

Bechard, M. J. 1982. Effect of vegetative cover on foraging site selection by Swainson’s

Hawk. The Condor 84:153-159.

Bechard, M. J., R. L. Knight, D. G. Smith, and R. E. Fitzner. 1990. Nest sites and habitats

of sympatric hawks (Buteo spp.) in Washington. Journal of Field Ornithology

84:159–170.

Bednarz, J. C., and J. J. Dinsmore. 1982. Nest-sites and habitat of Red-shouldered and

Red-tailed Hawks in Iowa. The Wilson Bulletin 94:31–45.

Benítez-López, A., R. Alkemade, and P. A. Verweij. 2010. The impacts of roads and

other infrastructure on mammal and bird populations: a meta-analysis. Biological

Conservation 143:1307–1316.

Boarman, W. I. 1993. When a native predator becomes a pest: a case study. Conservation

and Resource Management p. 186–201.

Boarman, W. I., M. A. Patten, R. J. Camp, and S. J. Collis. 2006. Ecology of a population

of subsidized predators: Common Ravens in the central Mojave Desert, California.

Journal of Arid Environments 67:248–261.

118

Bui, T.-V. D., Marzluff, and B. Bedrosian. 2010. Common Raven activity in relation to

land use in western Wyoming: implications for Greater sage-grouse reproductive

success. The Condor 112:65–78.

Burnham, K. P., and D. R. Anderson. 1998. Model selection and inference: a practical

information-theoretic approach. Springer. 382p.

Chalfoun, A. D., F. R. Thompson, and M. J. Ratnaswamy. 2002. Nest predators and

fragmentation: a review and meta-analysis. Conservation Biology 16:306–318.

Coates, P. S., and D. J. Delehanty. 2004. The effects of raven removal on sage grouse

nest success. Pages 17–20 Proceedings of the Vertebrate Pest Conference.

Coates, P. S., and D. J. Delehanty. 2010. Nest predation of Greater sage-grouse in

relation to microhabitat factors and predators. The Journal of Wildlife Management

74:240–248.

Dale, B. C., T. S. Wiens, and L. E. Hamilton. 2009. Abundance of three grassland

songbirds in an area of natural gas infill drilling in Alberta, Canada. Proceedings of

the Fourth International Partners in Flight Conference: Tundra to Tropics 194–204.

Davis, S. K., S. L. Jones, K. Dohms, and T. Holmes. 2012. Identification of Sprague’s

pipit nest predators. USGS Northern Prairie Wildlife Research Center. Paper 251.

Dechant, J. A., M. L. Sondreal, D. H. Johnson, L. D. Igl, C. M. Goldade, M. P.

Nenneman, and B. R. Euliss. 2002. Effects of management practices on grassland

birds: Northern Harrier. USGS Northern Prairie Wildlife Research Center No. 138.

Dijak, W. D., and F. R. Thompson. 2000. Landscape and edge effects on the distribution

of mammalian predators in Missouri. Journal of Wildlife Management 64:209–216.

119

Dinkins, J. B. 2013. Common Raven density and greater sage-grouse nesting success in

southern Wyoming: potential conservation and management implications (Thesis).

Utah State University.

Dion, N., K. A. Hobson, and S. Larivière. 2000. Interactive effects of vegetation and

predators on the success of natural and simulated nests of grassland songbirds. The

Condor 102:629–634.

Fiske, I., and R. Chandler. 2011. unmarked: An R package for fitting hierarchical models

of wildlife occurrence and abundance. Journal of Statistical Software 43:1–23.

Gilmer, D. S., and R. E. Stewart. 1984. Swainson’s Hawk nesting ecology in North

Dakota. The Condor 86:12-18.

Government of Saskatchewan. 2015. Oil and gas. Ministry of Economy and Resources.

Available at: http://www.economy.gov.sk.ca/oilandgas Accessed June 25, 2015.

Hall, T. R., W. E. Howard, and R. E. Marsh. 1981. Raptor use of artificial perches.

Wildlife Society Bulletin 9:296-298.

Hamilton, L. E., B. C. Dale, and C. A. Paszkowski. 2011. Effects of disturbance

associated with natural gas extraction on the occurrence of three grassland

songbirds. Avian Conservation and Ecology 6:7.

Herkert, J. R., S. A. Simpson, R. L. Westemeier, T. L. Esker and J. W. Walk. 1999.

Response of Northern Harriers and Short-eared Owls to grassland management in

Illinois. Journal of Wildlife Management 63:517-523.

Heske, E. J., S. K. Robinson, and J. D. Brawn. 1999. Predator activity and predation on

songbird nests on forest-field edges in east-central Illinois. Landscape Ecology

14:345–354.

120

Howe, K. B., P. S. Coates, and D. J. Delehanty. 2014. Selection of anthropogenic features

and vegetation characteristics by nesting Common Ravens in the sagebrush

ecosystem. The Condor 116:35–49.

Ingelfinger, F., and S. Anderson. 2004. Passerine response to roads associated with

natural gas extraction in a sagebrush steppe habitat. Western North American

Naturalist 64:385–395.

Janes, S. W. 1984. Influences of territory composition and interspecific competition on

Red-tailed Hawk reproductive success. Ecology 65:862–870.

Johnson, D. H. 2001. Habitat fragmentation effects on birds in grasslands and wetlands: a

critique of our knowledge. Great Plains Research 11:211-231.

Johnson, D. H. 2008. In defense of indices: the case of bird surveys. Journal of Wildlife

Management 72:857–868.

Johnson, D. H., and L. D. Igl. 2001. Area requirements of grassland birds: a regional

perspective. Auk 118:24–34.

Johnson, R. G., and S. A. Temple. 1990. Nest predation and brood parasitism of tallgrass

prairie birds. The Journal of Wildlife Management 54:106–111.

Kalyn-Bogard, H. J., and S. K. Davis. 2014. Grassland songbirds exhibit variable

responses to the proximity and density of natural gas wells. The Journal of Wildlife

Management 78:471–482.

Keough, H. L., and M. R. Conover. 2012. Breeding-site selection by ferruginous hawks

within Utah’s Uintah Basin. Journal of Raptor Research 46:378–388.

121

Knight, R. L., and J. Y. Kawashima. 1993. Responses of raven and Red-tailed Hawk

populations to linear right-of-ways. The Journal of Wildlife Management 57:266–

271.

Lariviere, S., L. R. Walton, and F. Messier. 1999. Selection by striped skunks (Mephitis

mephitis) of farmsteads and buildings as denning sites. American Midland Naturalist

142:96–101.

Ludlow, S. M., R. M. Brigham, and S. K. Davis. 2015. Oil and natural gas development

has mixed effects on the density and reproductive success of grassland songbirds.

The Condor 117:64–75.

Luttich, S., D. H. Rusch, E. C. Meslow, and L. B. Keith. 1970. Ecology of Red-tailed

Hawk predation in Alberta. Ecology 51:190-203.

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 J. A. Royle. 2005. Designing occupancy studies: general advice

and allocating survey effort. Journal of Applied Ecology 42:1105–1114.

Martin, J. W. 1987. Behavior and habitat use of breeding Northern Harriers in

southwestern Idaho. Journal of Raptor Research 21:57-66.

Martin, T. E. 1988. Processes organizing open-nesting bird assemblages: competition or

nest predation? Evolutionary Ecology 2:37-50.

Martin, T. E. 1993. Nest predation among vegetation layers and habitat types: revising

the dogmas. The American Naturalist 141:897–913.

122

McKercher, R. B., and B. Wolfe. 1986. Understanding western Canada's dominion land

survey system. Division of Extension and Community Relations report,

University of Saskatchewan, Saskatoon. 26 pp.

Phillips, M. L., W. R. Clark, M. A. Sovada, D. J. Horn, R. R. Koford, and R. J.

Greenwood. 2003. Predator selection of prairie landscape features and its relation to

duck nest success. The Journal of Wildlife Management 67:104–114.

Pietz, P. J., and D. A. Granfors. 2000. Identifying predators and fates of grassland

passerine nests using miniature video cameras. The Journal of Wildlife Management

64:71–87.

Preston, C. R. 1990. Distribution of raptor foraging in relation to prey biomass and

habitat structure. Condor 92:107–112.

Prugh, L. R., C. J. Stoner, C. W. Epps, W. T. Bean, W. J. Ripple, A. S. Laliberte, and J.

S. Brashares. 2009. The Rise of the mesopredator. BioScience 59:779–791.

Renfrew, R. B., and C. A. Ribic. 2003. Grassland passerine nest predators near pasture

edges identified on videotape. The Auk 120:371–383.

Renfrew, R. B., C. A. Ribic, and J. L. Nack. 2005. Edge avoidance by nesting grassland

birds: a futile strategy in a fragmented landscape. The Auk 122:618-636.

Ribic, C. A., M. Guzy, T. Anderson, D. Sample, and J. Nack. 2012. Bird productivity and

nest predation in agricultural grasslands. Video Surveillance of Nesting Birds.

Studies in Avian Biology No. 43. Cooper Ornithological Society. 240p.

Ricklefs, R. E. 1969. Nesting mortality in birds. Smithsonian Contributions to Zoology

No. 9.

123

Royle, J. A. 2004. N-mixture models for estimating population size from spatially

replicated counts. Biometrics 60:108–115.

Sargeant, A. B., R. J. Greenwood, M. A. Sovada, and T. L. Shaffer. 1993. Distribution

and abundance of predators that affect duck production-Prairie Pothole Region.

United States Department of the Interior, Fish and Wildlife Service No. 194.

Saskatchewan Research Council. 2015. FlySask.ca. Version 2.0. Saskatchewan

Geospatial Imagery Collaborative, http://www.flysask.ca, Accessed July 14, 2015.

Sauer, J. R., J. E. Hines, J. E. Fallon, K. L. Pardieck, D. J. Ziolkowski, Jr., and W. A.

Link. 2014. The North American breeding bird survey, results and analysis 1966 -

2013. Version 01.30.2015 USGS Patuxent Wildlife Research Center, Laurel, MD.

Schmutz, J. K. 1987. The effect of agriculture on Ferruginous and Swainson’s Hawks.

Journal of Range Management 40:438-440.

Souther, S., Tingley, M. W., Popescu, V. D., Hayman, D. T. S., Ryan, M. E., Graves, T.

A., Hartl, B., and Terrell, K. 2014. Biotic impacts of energy development from

shale: research priorities and knowledge gaps. Frontiers in Ecology and the

Environment 12:330-338.

Steenhof, K., M. N. Kochert, and J. A. Roppe. 1993. Nesting by raptors and common

ravens on electrical transmission line towers. The Journal of Wildlife Management

57:271–281.

Thompson, C. M., and E. M. Gese. 2013. Influence of vegetation structure on the small

mammal community in a shortgrass prairie ecosystem. Acta Theriologica 58:55–61.

Toland, B. R. 1985. Nest site selection, productivity, and food habits of Northern Harriers

in southwest Missouri. Natural Areas Journal 5:22-27.

124

Walk, J. W., and R. E. Warner. 1999. Effects of habitat area on the occurrence of

grassland birds in Illinois. The American Midland Naturalist 141:339-344.

Webb, W. C., W. I. Boarman, and J. T. Rotenberry. 2004. Common Raven juvenile

survival in a human-augmented landscape. The Condor 106:517–528.

Webb, S. L., C. V. Olson, M. R. Dzialak, S. M. Harju, J. B. Winstead, and D. Lockman.

2012. Landscape features and weather influence nest survival of a ground-nesting

bird of conservation concern, the Greater sage-grouse, in human-altered

environments. Ecological Processes 1:1–15.

Winter, M., D. H. Johnson, and J. Faaborg. 2000. Evidence for edge effects on multiple

levels in tallgrass prairie. Condor 102:256–266.

125

Table 3.1. Distribution of well densities in native and tame sample sites (n=243) in southeastern SK, in 2013-2014. ‘No Well’ = zero

wells, ‘Low Density’ = 1-4 wells, ‘Medium Density’ = 5-8 wells, ‘High Density’ = 9+ wells.

Habitat No Well Low Density Medium Density

High Density (max. well density)

Total

Native 28 32 29 23 (26)1 112 Tame 27 30 32 42 (48)1 131 TOTAL 55 62 61 65 243 1Maximum well density

126

Table 3.2. Mean percent of sample site area taken up by disturbance features associated with oil development, divided between well

density categories, in southeastern SK, 2013-2014. Cumulative Disturbance is the sum of all other disturbance features.

Disturbance Habitat No Well (± SE)

Low (± SE)

Medium (± SE)

High (± SE)

Oil Roads (%) native tame

0.2 ± 0.03 0.1 ± 0.03

0.3 ± 0.03 0.2 ± 0.04

0.6 ± 0.05 0.5 ± 0.05

1.0 ± 0.06 0.9 ± 0.07

Pipelines (%) native tame

0.2 ± 0.06 0.0 ± 0.04

0.1 ± 0.04 0.1 ± 0.04

0.3 ± 0.12 0.2 ± 0.07

0.4 ± 0.14 0.1 ± 0.03

Well Pads (%) native tame

1.0 ± 0.01 0.1 ± 0.05

0.2 ± 0.03 0.3 ± 0.08

0.5 ± 0.07 0.6 ± 0.09

0.8 ± 0.08 1.0 ± 0.10

Battery & Building Pads (%) native tame

1.0 ± 0.00 0.0 ± 0.02

0.1 ± 0.02 0.1 ± 0.05

0.2 ± 0.06 0.1 ± 0.04

0.3 ± 0.10 0.2 ± 0.04

Cumulative Disturbance (%) native tame

1.3 ± 0.27 2.3 ± 0.22

2.5 ± 0.31 2.8 ± 0.30

3.2 ± 0.35 3.9 ± 0.27

3.3 ± 0.31 4.8 ± 0.26

Power lines (m) native tame

370 ± 132 834 ± 193

1670 ± 223 1236 ± 196

2276 ± 312 2371 ± 279

3481 ± 332 2369 ± 211

127

Table 3.3. Correlations between amount of area disturbed in sample sites (n=243) by oil development features in southeastern SK,

2013-2014.

Battery & Building

Pads

Oil Roads Pipelines Power lines

Cumulative Disturbance

Well Density

Well Pads

Battery & Building Pads 1.00 0.38 0.25 0.29 0.43 0.37 0.24

Oil Roads 0.38 1.00 0.19 0.49 0.39 0.77 0.48

Pipelines 0.25 0.19 1.00 0.21 0.28 0.06 0.09

Power lines 0.29 0.49 0.21 1.00 0.47 0.39 0.25

Cumulative Disturbance 0.43 0.39 0.28 0.47 1.00 0.57 0.54 Well Density 0.37 0.77 0.06 0.39 0.57 1.00 0.71

Well Pads 0.24 0.48 0.09 0.25 0.54 0.71 1.00

128

Table 3.4. N-mixture occurrence models comparing and combining the most parsimonious landscape level disturbance models (area

disturbed by oil roads [oil roads], well pads [well pads], battery and oil building pads [bb pads], pipelines [pipelines], power line

length [power lines], and total sum of all disturbance features [cumulative disturbance]), plus habitat type [habitat], year [year], and

the observation only model [Detection] and the Null model in southeastern SK, 2013-2014. All disturbance models include the

observation model. All models are presented with the log likelihood score (LL), number of parameters (K), Akaike’s Information

Criterion score adjusted for small sample sizes (AICc), ∆AICc, and model weight (wi).

Model LL K AICc ∆AICc wi American Crow ahabitat*year habitat + year habitat Detection Null

-252.1 -254.8 -262.6 -265.1 -278.0

11 10

9 8 2

527.3 530.5 543.9 546.8 560.0

0.0 3.2

16.7 19.6 32.7

0.83 0.17 0.00 0.00 0.00

Black-billed Magpie Null

-150.9

2

305.8

0.0

1.00

Common Raven Detection Null

-107.9 -122.8

8 2

232.4 249.7

0.0

17.3

1.00 0.00

Northern Harrier awell # Detection

-140.5 -142.6

9 8

299.8 301.8

0.0 2.0

0.56 0.21

129

well # + habitat Null

-140.4 -150.6

10 2

301.8 305.3

2.1 5.6

0.20 0.03

Red-tailed Hawk Detection Null

-130.2 -146.0

9 2

279.2 296.0

0.0

16.8

1.00 0.00

Swainson’s Hawk Null

-161.8

2

327.6

0.0

1.00

aTop occurrence model for each species

130

Table 3.5. Parameter estimates of the top landscape disturbance N-mixture occurrence models of three corvid species and three hawk

species in southeastern SK, 2013-2014.

Parameter Estimate Standard Error

Lower 85% CI

Upper 85% CI

American Crow aIntercept habitat (native) ahabitat (tame) year (2013) year (2014) year(2013)*habitat(native) year(2013)*habitat(tame) year(2014)*habitat(native) ayear(2014*habitat(tame)

-0.686

0 1.899

0 -0.696

0 0 0

-1.834

0.330

0 0.657

0 0.550

0 0 0

0.844

-1.1612

0 0.9529

0 -1.4880

0 0 0

-3.0494

-0.2108

0 2.8451

0 0.0960

0 0 0

-0.6186

Black-billed Magpie aIntercept

-1.60

0.245

-1.9528

-1.2472

Common Raven Intercept

-0.16

1.12

-1.7728

1.4528

Northern Harrier aIntercept awell #

-0.876 -0.596

0.484 0.332

-1.5730 -1.0741

-0.1790 -0.1179

Red-tailed Hawk aIntercept

-1.34

0.339

-1.8282

-0.8518

131

Swainson’s Hawk Intercept

-0.597

0.480

-1.2882

0.0942

a85% confidence interval doesn’t include zero

132

Figure 3.1. Map of the study area, grass/pasture area, and 243 sample sites (with well densities) in southeastern SK, 2013-2014.

133

Figure 3.2. Typical sample site showing the 908 m and 100 m radius buffers, along with

the locations of the avian point count and oil wells.

134

Figure 3.3. Correlation between well density and total disturbed area in sample sites (n=243) in southeast SK, 2013-2014.

0 10 20 30 40

02

46

8

Well Density

Tota

l Dis

turb

ance

(%)

r = 0.57

135

Figure 3.4. Model-predicted mean occurrence (± 85% CI) of Northern Harriers varied by well density in southeastern SK, 2013-2014.

136

Figure 3.5. Model-predicted mean occurrence (± 85% CI) of American Crows varied by habitat type and year in southeastern SK,

2013-2014.

137

4. GENERAL CONCLUSIONS

I examined the influence of oil development on the abundance of grassland

songbirds and the occurrence of avian predators. It is evident from my results that oil

wells and the disturbance associated with them influence grassland songbird abundance

and Northern Harrier occurrence. While the particular effects of oil development are

somewhat species specific, I found a general overall trend that oil development

negatively influences abundance, particularly for grassland specialists. At least seven

species experienced reduced abundance closer to oil well or in areas with high well-

density. Only two species, Brown-headed Cowbird and Vesper Sparrow, had increased

abundance in the presence of oil wells, while Clay-colored Sparrow and Horned Lark did

not respond to the presence of oil wells. No species exhibited greater abundance in areas

with higher well density. Furthermore, the amount of landscape disturbance associated

with oil development in a 259 ha area influences grassland songbird abundance. Three

species had decreased abundance in areas with higher cumulative disturbance, while two

species exhibited reduced abundance when the amount of land converted to well pads or

oil roads increased. Grasshopper Sparrow showed increased abundance as oil battery and

building area increased, although, as cumulative disturbance increased beyond a

threshold of 3% of the landscape Grasshopper Sparrow abundance was reduced.

I also found evidence that the type of grassland habitat plays a role in how some

species respond to the effects of oil development. The abundance of two grassland

specialists and two generalist species was lowest in planted grasslands where oil

development was present compared to native pastures. However, Bobolink abundance

was greatest in planted grasslands with oil development compared to native pastures.

138

Thus, factors that determine good habitat quality in the presence of oil development are

not the same for all species.

Oil development influences vegetation structure, which could in turn affect

grassland songbird abundance. However, I accounted for vegetation effects in my

models so there must be other factors associated with oil development influencing bird

abundance that extend beyond changes to vegetation. What the specific mechanisms are

that influence abundance remains unclear. However, the negative effects of oil

development on grassland songbirds are likely a reflection of additional stresses on an

already stressed habitat. The fact that birds experience reduced abundance in the

presence of oil development is an indication that habitat quality may be further reduced

by the pressures of oil development.

My study suggests that oil development may have a stronger influence on

grassland songbirds than other types of energy development (e.g. natural gas, wind

energy). Mine is one of the first studies to examine the specific effects of oil well

proximity, density, and cumulative disturbance on grassland songbirds. My results are

consistent with the hypothesis that oil development reduces the quality of habitat for

grassland specialists such as Baird’s Sparrow and Sprague’s Pipit. However, my results

show that oil development also negatively affects species that prefer planted pastures (i.e.

Bobolink) and generalist species such as Savannah Sparrow, even though other types of

energy development typically positively influence Savannah Sparrow.

My results provide evidence that Northern Harrier occurrence is negatively

influenced by oil development. Northern Harrier experienced reduced occurrence as well

density increased. Northern Harrier occurrence is possibly influenced by habitat

139

fragmentation caused by oil development since they are known to be area sensitive.

However, the occurrence of other potential songbird predators in my study were not

affected by oil development. The occurrence of buteos and corvids may not have been

influenced by oil development because of an abundance of perch and nest sites not

associated with oil development. My results also show that grassland habitat type (native

and planted) did not influence the occurrence of most potential avian predators.

American Crow was the only species influenced by habitat type and occurrence was

greater in planted grasslands in one of the two years. These results suggest that oil

development and grassland habitat type are not important factors in determining the

occurrence of buteos and most corvid species in my study, which in turn means that this

is not likely the mechanism leading to reduced habitat quality for grassland songbirds.

Based on my results, several recommendations can be made to land managers faced

with decisions regarding grassland songbird conservation in the midst of oil

development:

1) Efforts should be taken to limit well density and associated cumulative

disturbance in grassland habitat. Several species exhibited reduced abundance or

occurrence as well density increased, and no species were more abundant in areas

with increased well densities. Also, several songbird species were less abundant

closer to wells, and as well density increases the distance between wells will

decrease so limiting well density will also benefit species that avoid areas near

wells.

2) Wells should be placed as close together as possible, preferably on the same pads.

This will serve to limit the effects of well proximity on grassland birds, as well as

140

limiting the amount of disturbance from oil roads and other disturbance features.

Localizing disturbance and limiting wells to smaller areas is possible with today’s

directional drilling technology.

3) Oil development should be avoided on native grasslands as much as possible.

Several species sensitive to oil development are more abundant on native

grasslands, likely because it represents better quality habitat. Reducing oil

development on native grasslands will preserve the functional quality of native

grasslands for these species.

4) Efforts need to be made to limit the impact to surrounding vegetation when oil

wells are drilled.

141

APPENDIX A – SONGBIRD OBSERVATION MODELS

It is essential to estimate and incorporate detection probabilities in bird abundance

models. Detection probabilities account for birds present but undetected because they did

not elicit a cue to be recorded. Thus, repeat visits by multiple observers allow for

detection estimates and observer biases to be accounted for in estimating abundance

(MacKenzie et al. 2002, MacKenzie and Royle 2005, Johnson 2008).

The N-mixture abundance models developed by Royle (2004) have two ‘sides’:

an observation model, which uses a Binomial distribution to estimate detection

probability, and an abundance model, which is directly informed by the observation

model. In developing my abundance models, I first selected the most parsimonious

observation model for each grassland songbird species.

I used four detection parameters in the observation models: ordinal date of survey,

time of survey, wind speed during survey, and observer. There were seven observers

over the course of the study, and one observer performed surveys in 2013 and 2014. The

other six observers only conducted surveys in one of the two study years. I also included

two site-specific parameters in the observation models (well density, distance to nearest

well) in case the presence, movement, or noise from oil wells influenced bird detection. I

used abundance counts of singing males within a 100 m radius for all species, except

Brown-headed Cowbirds (Molothrus ater) where I used the total counts of both males

and females. I used Akaike’s Information Criterion adjusted for small sample sizes

(AICc) for model comparison and selection (Burnham and Anderson 1998), and I used

85% confidence limits to identify uninformative parameters (Arnold 2010) and to

142

determine the strength of effect of each informative parameter. I compared the

interactive and additive effects of year for each detection parameter to determine if I

could combine the two years of data together. I then created a global model by

comparing each detection parameter to the Null model and retaining the detection

parameters that performed better than the Null model. I fitted all subset models of the

global model and selected the top observation model to be used as the detection

component of the abundance model for each species. I considered the top fitted model to

be the one with the highest AIC weight (wi). If there were competing models (within

∆AICc < 2 and the same or fewer parameters as the model with the highest AIC weight), I

selected the most parsimonious model as the top model. I used this method for model

selection in all further model analyses.

The observation model selection for most species was straightforward. I selected

the top ranked model based on the lowest AICc scores for 10 grassland songbird species

(Table A.1). All detection variables in the selected observation models did not include

zero in their 85% confidence intervals (Table A.2). However, for Bobolink I chose the

second-ranked observation model (Table A.1) because the 85% confidence interval for

well distance included zero in the top ranked model. Also, the relative variable

importance of well distance was low in all subsets of Bobolink observation models (A.3).

These observation models were included in all vegetation, oil well, and disturbance

model analyses.

LITERATURE CITED

Arnold, T. W. 2010. Uninformative parameters and model selection using Akaike’s

Information Criterion. The Journal of Wildlife Management 74:1175–1178.

143

Burnham, K. P., and D. R. Anderson. 1998. Model Selection and Inference: A Practical

Information-theoretic Approach. Springer. 353p.

Johnson, D. H. 2008. In Defense of indices: The case of bird surveys. Journal of Wildlife

Management 72:857–868.

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 J. A. Royle. 2005. Designing occupancy studies: General advice

and allocating survey effort. Journal of Applied Ecology 42:1105–1114.

Royle, J. A. 2004. N-mixture models for estimating population size from spatially

replicated counts. Biometrics 60:108–115.

144

Table A.1. N-mixture models comparing the most parsimonious observation models for eleven species of grassland songbirds in

southeastern SK, 2013-2014. Detection covariates include: observer, ordinal date (date), wind speed (wind), time of day (time),

distance to nearest well (well dist), and well density (density). Models with weights (wi) that sum to 0.99 are shown, plus the Null

model. All models are presented with the log likelihood score (LL), number of parameters (K), Akaike’s Information Criterion score

adjusted for small sample sizes (AICc), ∆AICc, and model weight (wi).

Model LL K AICc ∆AICc wi Baird’s Sparrow aobserver + date + wind + well dist + density observer + date + wind + time + well dist + density observer + date + time + well dist + density observer + date + wind + well dist Null

-1148.0 -1147.6 -1152.0 -1153.7 -1248.8

12 13 12 11 2

2320.7 2322.1 2328.7 2329.9 2501.6

0.0 1.4 8.0 9.2

180.9

0.63 0.34 0.01 0.01 0.00

Brown-headed Cowbird aobserver + date + time + well dist observer + date + well dist Null

-3188.8 -3193.0 -3283.6

11 10 2

6400.1 6406.6 6571.2

0.0 6.5

171.1

0.96 0.04 0.00

Bobolink observer + well dist aobserver well dist Null

-1467.6 -1469.4 -1505.9 -1507.3

9 8 3 2

2953.6 2955.0 3017.8 3018.5

0.0 1.4

64.2 64.9

0.68 0.32 0.00 0.00

Chestnut-collared Longspur

145

aobserver + date + wind + density observer + date + wind Null

-495.9 -501.3 -572.9

11 10 2

1014.4 1023.1 1149.9

0.0 8.7

135.5

0.98 0.01 0.00

Clay-colored Sparrow aobserver + date + wind + time observer + wind + time observer + wind + date Null

-2301.5 -2303.5 -2306.6 -2455.8

11 10 10 2

4625.6 4627.4 4633.6 4915.5

0.0 1.8 8.0

289.9

0.71 0.27 0.01 0.00

Grasshopper Sparrow aobserver + date + time + density observer + date + time + wind + density observer + date + time observer + date + time + wind Null

-920.3 -920.1 -924.0 -924.0

-1122.0

11 12 10 11 2

1863.3 1864.9 1868.5 1870.5 2248.0

0.0 1.6 5.3 7.3

384.8

0.64 0.30 0.04 0.01 0.00

Horned Lark aobserver + well dist observer + well dist + density Null

-578.7 -578.7 -609.3

9

10 2

1175.8 1177.8 1222.6

0.0 2.0

46.8

0.72 0.27 0.00

Savannah Sparrow aobserver + date + wind + time + density observer + date + wind + time + density + well dist observer + date + wind + time + well dist observer + date + wind + time observer + date + time + density observer + date + time + density + well dist observer + date + time + well dist observer + date + time

-2626.2 -2625.8 -2626.9 -2628.8 -2628.9 -2628.6 -2629.8 -2631.2

12 13 12 11 11 12 11 10

5277.1 5278.3 5278.4 5280.1 5280.4 5281.9 5282.1 5282.9

0.0 1.2 1.3 3.0 3.2 4.8 5.0 5.8

0.37 0.21 0.19 0.08 0.07 0.03 0.03 0.01

146

Null -2699.2 2 5402.5 125.4 0.00

Sprague’s Pipit aobserver + wind + density observer + density observer + wind observer Null

-587.5 -589.4 -590.4 -591.6 -634.0

10 9 9 8 2

1195.5 1197.2 1199.2 1199.5 1272.0

0.0 1.7 3.7 4.1

76.6

0.60 0.24 0.09 0.07 0.00

Vesper Sparrow aobserver + wind + density observer + density observer + wind Null

-1119.8 -1122.5 -1123.1 -1162.0

10 9 9 2

2260.1 2263.3 2264.6 2328.0

0.0 3.2 4.5

67.9

0.77 0.14 0.08 0.00

Western Meadowlark aobserver + date + wind + time observer + date + time Null

-1386.9 -1390.1 -1514.1

11 10 2

2796.4 2800.6 3032.3

0.0 4.2

235.9

0.89 0.10 0.00

aMost parsimonious observation model for each species.

147

Table A.2. Parameter estimates of the top observation models of eleven grassland songbird species in southeastern SK, 2013-2014.

Parameter Estimate Standard Error

Lower 85% CI

Upper 85% CI

Baird’s Sparrow Intercept awell distance adensity obs1 aobs2 aobs3 aobs4 aobs5 aobs6 aobs7 adate awind

-0.0085 0.3920

-0.3909 0

-1.1409 -1.9991 -1.3186 -1.8173 -1.6365 -1.4632 -0.2707 -0.1941

0.1748 0.0865 0.1205

0 0.1781 0.2493 0.1656 0.2377 0.1920 0.2359 0.0751 0.0584

-0.2602 0.2674

-0.5644 0

-1.3974 -2.3581 -1.5571 -2.1596 -1.9130 -1.8029 -0.3788 -0.2782

0.2432 0.5166

-0.2174 0

-0.8844 -1.6401 -1.0801 -1.4750 -1.3600 -1.1235 -0.1626 -0.1100

Brown-headed Cowbird aIntercept adate atime obs1 aobs2 obs3 aobs4 aobs5 obs6 aobs7

awell distance

-2.2462 -0.1401 -0.0647

0 0.2648

-0.0139 0.4787 0.5123

-0.0662 0.5793

-0.1172

0.1403 0.0268 0.0222

0 0.0924 0.0999 0.0807 0.0924 0.0984 0.0919 0.0278

-2.4482 -0.1787 -0.0967

0 0.1317

-0.1578 0.3625 0.3792

-0.2079 0.4470

-0.1572

-2.0442 -0.1015 -0.0327

0 0.3979 0.1300 0.5949 0.6454 0.0755 0.7116

-0.0772

148

Bobolink Intercept obs1 aobs2 aobs3 aobs4 obs5 aobs6 obs7

-0.0719

0 -0.9747 -0.8166 -0.2697 -0.2681 -1.0187 0.1474

0.164

0 0.199 0.207 0.167 0.202 0.200 0.204

-0.3081

0 -1.2613 -1.1147 -0.5102 -0.5590 -1.3067 -0.1464

0.1643

0 -0.6881 -0.5185 -0.0292 0.0228

-0.7307 0.4412

Chestnut-collared Longspur aIntercept obs1 aobs2 aobs3 aobs4 aobs5 obs6 aobs7 adate awind adensity

-1.835

0 1.891

-0.987 0.464

-0.802 -0.344 -1.610 -0.878 -0.555 -0.592

0.270

0 0.329 0.493 0.314 0.454 0.372 0.609 0.151 0.123 0.184

-2.2238

0 1.4172

-1.6969 0.0118

-1.4558 -0.8797 -2.4870 -1.0954 -0.7321 -0.8570

-1.4462

0 2.3648

-0.2771 0.9162

-0.1482 0.1917

-0.7330 -0.6606 -0.3779 -0.3270

Clay-colored Sparrow aIntercept adate awind atime obs1 obs2

-0.9541 0.1012

-0.2412 -0.1094

0 0.0774

0.1059 0.0512 0.0384 0.0345

0 0.1248

-1.1066 0.0275

-0.2965 -0.1591

0 -0.1023

-0.8016 0.1749

-0.1859 -0.0597

0 0.2571

149

aobs3 aobs4 aobs5 aobs6 aobs7

-0.2176 1.0642 0.6233 0.6861 1.0062

0.1421 0.1142 0.1372 0.1221 0.1421

-0.4222 0.8998 0.4257 0.5103 0.8016

-0.0130 1.2286 0.8209 0.8619 1.2108

Grasshopper Sparrow aIntercept adate atime obs1 aobs2 aobs3 aobs4 aobs5 aobs6 aobs7 adensity

-4.375 0.580

-0.380 0

1.781 2.800 3.491 1.000 4.017 3.622

-0.312

0.4648 0.0907 0.0716

0 0.4994 0.4990 0.4695 0.5745 0.4770 0.4968 0.1031

-5.0443 0.4494

-0.4831 0

1.0619 2.0814 2.8149 0.1727 3.3301 2.9066

-0.4605

-3.7057 0.7106

-0.2769 0

2.5001 3.5186 4.1671 1.8273 4.7039 4.3374

-0.1635

Horned Lark aIntercept obs1 obs2 aobs3 aobs4 aobs5 obs6 obs7 awell distance

-1.380

0 -0.289 -1.350 0.600

-0.568 0.00002

0.095 -0.609

0.269

0 0.341 0.450 0.282 0.377 0.327 0.347 0.131

-1.7674

0 -0.7800 -1.9980 0.1939

-1.1109 -0.4709 -0.4047 -0.7976

-0.9926

0 0.2020

-0.7020 1.0061

-0.0251 0.4709 0.5947

-0.4204 Savannah Sparrow aIntercept

-0.9865

0.1019

-1.1332

-0.8398

150

adate awind atime adensity obs1 aobs2 aobs3 aobs4 aobs5 aobs6 aobs7

0.2038 -0.0724 -0.1741 -0.1061

0 0.3118 0.5406 0.7087 0.5801 0.5092 0.6500

0.0457 0.0315 0.0293 0.0457

0 0.1019 0.1182 0.0947 0.1181 0.1015 0.1197

0.1380 -0.1178 -0.2163 -0.1719

0 0.1651 0.3704 0.5723 0.4100 0.3630 0.4776

0.2696 -0.0270 -0.1319 -0.0403

0 0.4585 0.7108 0.8451 0.7502 0.6554 0.8224

Sprague’s Pipit aIntercept obs1 aobs2 aobs3 obs4 obs5 aobs6 aobs7

awind adensity

-0.631

0 -1.327 -0.559 0.202 0.216

-0.593 -3.777 -0.196 -0.442

0.230

0 0.342 0.365 0.258 0.359 0.269 1.034 0.101 0.161

-0.9622

0 -1.8195 -1.0846 -0.1695 -0.3010 -0.9804 -5.2660 -0.3414 -0.6738

-0.2998

0 -0.8354 -0.0334 0.5735 0.7330

-0.2056 -2.2880 -0.0506 -0.2102

Vesper Sparrow aIntercept awind obs1 obs2 aobs3 obs4

-1.1529 -0.1439

0 0.1144

-1.3148 0.2528

0.1994 0.0627

0 0.2195 0.2917 0.1934

-1.4400 -0.2342

0 -0.2017 -1.7348 -0.0257

-0.8658 -0.0536

0 0.4305

-0.8948 0.5313

151

obs5 obs6 aobs7 adensity

0.3132 0.0592 0.5369 0.1921

0.2288 0.2208 0.2283 0.0748

-0.0163 -0.2588 0.2081 0.0844

0.6427 0.3772 0.8657 0.2998

Western Meadowlark aIntercept adate awind atime obs1 aobs2 aobs3 aobs4 aobs5 aobs6 aobs7

-0.577 -0.306 -0.129 -0.173

0 -1.932 -1.561 -1.183 -1.444 -1.634 -0.583

0.2481 0.0573 0.0513 0.0499

0 0.2030 0.1960 0.1477 0.1905 0.1863 0.1669

-0.9343 -0.3885 -0.2029 -0.2449

0 -2.2243 -1.8432 -1.3957 -1.7183 -1.9023 -0.8233

-0.2197 -0.2235 -0.0551 -0.1011

0 -1.6397 -1.2788 -0.9703 -1.1697 -1.3657 -0.3427

a85% confidence interval doesn’t include zero

152

Table A.3. Relative importance of each detection variable in selecting the top

observation model for eleven grassland songbird species in southeastern SK, 2013-2014.

Detection Parameters Relative Variable Importance Baird’s Sparrow date observer well distance well density wind time

1.00 1.00 1.00 0.99 0.98 0.35

Brown-headed Cowbird date observer well distance time

1.00 1.00 1.00 0.96

Bobolink observer well distance

1.00 0.68

Chestnut-collared Longspur observer date wind well density

1.00 1.00 1.00 0.99

Clay-colored Sparrow wind observer time date

1.00 1.00 0.98 0.72

Grasshopper Sparrow date time observer well density wind

1.00 1.00 1.00 0.94 0.32

Horned Lark well distance observer well density

1.00 1.00 0.27

153

Savannah Sparrow date time observer wind well density well distance

1.00 1.00 1.00 0.85 0.68 0.47

Sprague’s Pipit observer well density wind

1.00 0.84 0.69

Vesper Sparrow observer well density wind

1.00 0.91 0.85

Western Meadowlark date time observer wind

1.00 1.00 1.00 0.90

154

APPENDIX B – SONGBIRD VEGETATION MODELS

I used N-mixture abundance models to determine the top vegetation model for

each grassland songbird species. I considered eight vegetation parameters: native grass

cover, planted grass cover, forb cover, bare ground cover, distance to nearest shrub, litter

depth, vegetation height, and visual obstruction (Robel pole). I first evaluated the

interactive and additive effects of year with each vegetation parameter to determine if I

could combine the two years of data together. I then compared the linear and quadratic

effect of each vegetation parameter to the observation only model and selected the linear

or quadratic relationships that performed better than the observation model for each

species. I fitted all subsets of the remaining vegetation parameters and selected the top

vegetation model for each species.

The top ranked vegetation model based on the lowest AICc score was selected for

Brown-headed Cowbird, Bobolink, Chestnut-collared Longspur, Horned Lark, and

Savannah Sparrow (Table B.1). Competing models were selected as the best vegetation

models for Baird’s Sparrow, Clay-colored Sparrow, Sprague’s Pipit, Vesper Sparrow,

and Western Meadowlark (Table B.1). These models were chosen because the top

ranked models included uninformative variables.

Grasshopper Sparrow was the only species that had a significant interaction

between year and vegetation variables. Grasshopper Sparrow abundance had a positive

relationship with litter depth in 2013 and a negative relationship in 2014 (Fig. B.1). Thus,

Grasshopper Sparrow abundance in 2013 and 2014 was analyzed separately.

155

Table B.1. N-mixture abundance models comparing the most parsimonious vegetation models for eleven grassland songbird species

in southeastern SK, 2013-2014. Model variables include: native grass cover (native), planted grass cover (tame), forb cover (forbs),

bare ground cover (bare), litter depth (litter), distance to nearest shrub (shrubs), visual obstruction (robel), and vegetation height

(height). Models with weights (wi) that sum to 0.90 are shown, plus the observation (Detection) and Null models. All vegetation

models include the observation model (except the Null model). All models are presented with the log likelihood score (LL), number

of parameters (K), Akaike’s Information Criterion score adjusted for small sample sizes (AICc), ∆AICc, and model weight (wi).

Model LL K AICc ∆AICc wi Baird’s Sparrow native + forbs2 + shrubs2 + robel2 + litter native + tame + forbs2 + shrubs2 + robel2 + litter native + forbs2 + shrubs2 + robel2 + litter + bare anative + shrubs2 + forbs2 + litter native + tame + forbs2 + robel2 + litter native + tame + shrubs2 + robel2 + litter native + forbs2 + litter Detection Null

-1084.8 -1084.1 -1084.1 -1086.6 -1086.2 -1086.3 -1088.5 -1148.0 -1248.8

17 18 18 16 17 17 15 12 2

2205.0 2205.7 2205.7 2206.3 2207.7 2207.9 2208.0 2320.7 2501.6

0.0 0.8 0.8 1.3 2.8 2.9 3.0

115.7 296.7

0.24 0.17 0.16 0.12 0.06 0.06 0.05 0.00 0.00

Brown-headed Cowbird arobel2 Detection Null

-3181.3 -3188.8 -3283.6

12 11 2

6387.3 6400.1 6571.2

0.0

12.8 183.9

1.00 0.00 0.00

Bobolink

156

atame + shrubs2 + height native + tame + shrubs2 + height tame + shrubs2 + height + litter Detection Null

-1421.3 -1420.9 -1421.3 -1469.4 -1507.3

11 12 12 8 2

2865.1 2866.4 2867.2 2955.0 3018.5

0.0 1.3 2.1

89.9 153.4

0.48 0.26 0.17 0.00 0.00

Chestnut-collared Longspur atame + shrubs2 + forbs2 + litter tame + forbs2 + shrubs2 + robel + litter native + tame + forbs2 + shrubs2 + robel + litter native + tame + forbs2 + shrubs2 + litter tame + forbs2 + litter native + tame + forbs2 + robel + litter native + tame + forbs2 + litter Detection Null

-426.0 -425.5 -425.7 -424.7 -428.4 -426.5 -427.7 -495.9 -527.9

15 16 16 17 14 16 15 11 2

883.0 884.1 884.5 884.6 885.7 886.1 886.5

1014.4 1149.9

0.0 1.1 1.5 1.6 2.6 3.1 3.4

131.4 266.9

0.27 0.17 0.14 0.14 0.07 0.06 0.05 0.00 0.00

Clay-colored Sparrow native + shrubs + robel + litter tame + shrubs + robel + litter ashrubs + robel + litter tame + bare + shrub + robel + litter native + tame + shrub + robel + litter native + bare + shrub + robel + litter Detection Null

-2249.3 -2249.4 -2250.9 -2249.1 -2249.2 -2249.3 -2296.7 -2455.8

15 15 14 16 16 16 12 2

4529.7 4529.9 4530.6 4531.4 4531.6 4531.8 4618.0 4915.5

0.0 0.2 0.9 1.7 1.9 2.1

88.3 385.8

0.24 0.23 0.15 0.10 0.09 0.08 0.00 0.00

Grasshopper Sparrow 2013 arobel2 shrubs + robel2 shrubs

-403.5 -406.8 -409.9

10 9 9

828.0 832.4 838.6

0.0 4.4

10.6

0.40 0.37 0.13

157

Detection Null

-412.1 -482.1

8 2

840.8 968.3

12.8 140.3

0.10 0.00

Grasshopper Sparrow 2014 shrubs + robel + litter ashrubs + litter Detection Null

-395.6 -397.8 -412.1 -482.1

11 10 8 2

814.3 816.5 840.8 968.3

0.0 2.2

26.4 154.0

0.73 0.23 0.00 0.00

Horned Lark atame + robel + litter native + robel + bare + litter native + +tame + robel + litter tame + robel + bare + litter Detection Null

-547.7 -547.2 -547.5 -547.7 -578.7 -609.3

12 13 13 13 9 2

1120.1 1121.2 1121.9 1122.2 1175.8 1222.6

0.0 1.0 1.7 2.1

55.6 102.4

0.37 0.22 0.16 0.13 0.00 0.00

Savannah Sparrow bare + shrubs + robel2 + forbs abare + shrubs + robel2 bare + tame2 + shrubs + robel2 bare + tame2 + shrubs + robel + forbs Detection Null

-2610.8 -2612.4 -2611.6 -2610.6 -2626.2 -2699.2

16 15 16 17 12 2

5254.8 5255.9 5256.4 5256.6 5277.1 5402.5

0.0 1.1 1.6 1.8

22.3 147.7

0.41 0.24 0.19 0.17 0.00 0.00

Sprague’s Pipit native + tame + robel + litter native + tame + shrubs + robel + litter native + tame + forbs2 + robel + litter native + tame +forbs2 + shrubs + robel + litter anative + robel + litter

-538.0 -537.0 -537.2 -536.3 -539.8

14 15 15 16 13

1104.8 1105.1 1105.4 1105.8 1106.3

0.0 0.3 0.6 1.0 1.5

0.19 0.17 0.15 0.13 0.08

158

native + shrubs + robel + litter native + forbs2 + robel + litter native + forbs2 + shrubs + robel + litter tame + forbs2 + shrubs + robel + litter Detection Null

-538.8 -539.5 -538.6 -538.7 -587.5 -634.0

14 14 15 15 10 2

1106.5 1107.9 1108.2 1108.4 1195.5 1272.0

1.7 3.1 3.4 3.6

90.7 167.2

0.08 0.04 0.04 0.03 0.00 0.00

Vesper Sparrow native2 + tame2 + shrubs + robel native2 + tame2 + forbs + shrubs + robel native2 + tame2 + shrubs + robel + litter native2 + tame2 + shrubs + robel + bare native2 + tame2 + forbs + robel native2 + tame2 + forbs +shrubs + robel + litter native2 + tame2 + robel + litter anative2 + tame2 + robel native2 + tame2 + forbs + robel + litter native2 + tame2 + forbs + shrubs + robel + bare native2 + tame2 + forbs + shrubs + robel + litter + bare native2 + tame2 + forbs + robel + bare tame2 + shrubs + forbs + robel native2 + tame2 + robel + bare Detection Null

-1099.2 -1098.3 -1098.5 -1098.6 -1099.7 -1100.9 -1099.9 -1097.8 1099.0 1098.0 1098.1

-1100.7 -1100.8 -1099.7 -1119.8 -1162.0

14 15 15 15 14 13 14 16 15 16 16

14 14 15 10 2

2227.2 2227.7 2228.1 2228.3 2228.3 2228.6 2228.7 2228.9 2229.0 2229.2 2229.4

2230.3 2230.4 2230.4 2260.1 2328.0

0.0 0.5 0.9 1.1 1.1 1.4 1.5 1.7 1.8 2.0 2.2

3.1 3.2 3.2

32.9 100.8

0.14 0.11 0.09 0.08 0.08 0.07 0.06 0.06 0.05 0.05 0.05

0.03 0.03 0.03 0.00 0.00

Western Meadowlark native + forbs + robel + litter2 anative + robel + litter2

native + forbs + litter2

native + tame + forbs + robel +litter2

-1372.4 -1373.9 -1374.2 -1372.4

15 14 14 16

2775.8 2776.7 2777.2 2777.9

0.0 0.9 1.4 2.1

0.27 0.16 0.13 0.10

159

native + tame + robel + litter2

native + tame + forbs + litter2

tame + forbs + litter2

tame + forbs + robel + litter2 Detection Null

-1373.5 -1375.1 -1374.1 -1374.2 -1386.9 -1514.1

15 14 15 15 11 2

2777.9 2779.1 2779.1 2779.4 2796.4 3032.3

2.1 3.3 3.3 3.6

20.6 256.5

0.09 0.05 0.05 0.05 0.00 0.00

aMost parsimonious vegetation model for each species

160

Table B.2. Parameter estimates of the top vegetation models of eleven grassland songbird species in southeastern SK, 2013-2014.

Parameter Estimate Standard Error

Lower 85% CI

Upper 85% CI

Baird’s Sparrow Intercept anative ashrubs2 aforbs2 alitter

0.174 0.486

-0.146 -0.162 -0.170

0.1253 0.0609 0.0767 0.0722 0.0648

-0.0064 0.3983

-0.2564 -0.2660 -0.2633

0.3544 0.5737

-0.0356 -0.0580 -0.0767

Brown-headed Cowbird aIntercept arobel2

2.3364 0.0433

0.1104 0.0104

2.1774 0.0283

2.4954 0.0583

Bobolink Intercept atame ashrubs2 aveg height

-0.137 0.264 0.131 0.297

0.0806 0.0508 0.0456 0.0452

-0.2531 0.1908 0.0653 0.2319

-0.0209 0.3372 0.1967 0.3621

Chestnut-collared Longspur aIntercept atame ashrubs2 aforbs2 alitter

-1.216 -1.363 -0.326 -0.488 -0.412

0.274 0.216 0.162 0.208 0.143

-1.6106 -1.6740 -0.5593 -0.7875 -0.6179

-0.8214 -1.0520 -0.0927 -0.1885 -0.2061

Clay-colored Sparrow aIntercept

0.9196

0.0527

0.8437

0.9955

161

ashrubs2 arobel alitter

-0.3091 0.1245 0.0786

0.0383 0.0334 0.0300

-0.3643 0.0764 0.0354

-0.2539 0.1726 0.1218

Grasshopper Sparrow 2013 aIntercept arobel2

-0.396 -0.139

0.1325 0.0773

-0.5868 -0.2503

-0.2052 -0.0277

Grasshopper Sparrow 2014 aIntercept ashrubs alitter

-0.678 0.229

-0.594

0.1290 0.0889 0.1595

-0.8638 0.1010

-0.8237

-0.4922 0.3570

-0.3543

Horned Lark aIntercept atame arobel alitter

-0.986 -0.309 -0.491 -0.460

0.151 0.102 0.126 0.147

-1.2034 -0.4559 -0.6724 -0.6717

-0.7686 -0.1621 -0.3096 -0.2483

Savannah Sparrow aIntercept abare ground ashrubs arobel2

1.3721

-0.1086 0.0801

-0.0567

0.0635 0.0301 0.0277 0.0183

1.2807

-0.1519 0.0402

-0.0831

1.4635

-0.0653 0.1200

-0.0303

Sprague’s Pipit aIntercept anative arobel alitter

-1.294 0.712

-0.432 -0.361

0.1466 0.0908 0.1281 0.1219

-1.5051 0.5812

-0.6165 -0.5365

-1.0829 0.8428

-0.2475 -0.1855

162

Vesper Sparrow aIntercept anative2 atame2 arobel

0.482

-0.219 -0.275 -0.209

0.1316 0.0826 0.0817 0.0618

0.2925

-0.3379 -0.3926 -0.2980

0.6715

-0.1001 -0.1574 -0.1200

Western Meadowlark aIntercept anative arobel alitter2

0.8861 0.1579

-0.1239 -0.0495

0.1625 0.0440 0.0524 0.0225

0.6521 0.0945

-0.1994 -0.0819

1.1201 0.2213

-0.0484 -0.0171

a85% confidence interval doesn’t include zero

163

Table B.3. Relative importance of each vegetation variable in selecting the top vegetation

model for eleven grassland songbird species in southeastern SK, 2013-2014.

Vegetation parameters Relative Variable Importance Baird’s Sparrow native litter forbs shrubs robel tame bare ground

0.97 0.93 0.89 0.79 0.75 0.47 0.28

Brown-headed Cowbird robel

1.00

Bobolink veg height tame shrubs native litter

1.00 1.00 0.96 0.35 0.27

Chestnut-collared Longspur tame forbs litter shrubs robel native

1.00 0.97 0.97 0.77 0.46 0.42

Clay-colored Sparrow robel shrubs litter native tame bare ground

1.00 1.00 0.90 0.48 0.46 0.30

Grasshopper Sparrow 2013 robel shrubs

0.77 0.50

Grasshopper Sparrow 2014 litter

1.00

164

shrubs robel

0.96 0.75

Horned Lark litter robel tame native bare ground

1.00 1.00 0.67 0.54 0.41

Savannah Sparrow robel bare ground shrubs forbs tame

0.98 0.97 0.95 0.58 0.39

Sprague’s Pipit litter robel native tame shrubs forbs

0.97 0.96 0.92 0.76 0.50 0.45

Vesper Sparrow robel tame native shrubs forbs litter bare ground

0.99 0.98 0.86 0.66 0.49 0.41 0.30

Western Meadowlark litter native robel forbs tame

0.95 0.89 0.73 0.68 0.37

165

APPENDIX C – SONGBIRD OIL WELL MODELS

I used N-mixture abundance models to determine the top oil well model for each

grassland songbird species. I considered three oil well parameters: distance to nearest

well, well density, and well activity. Well activity was determined by whether the well

was an actively producing well (‘active’), a well no longer producing oil (‘abandoned’),

or no well present (‘none’). I first evaluated the interactive and additive effects of year

with each oil well parameter to determine if I could combine the two years of data

together. I then compared the linear and quadratic effect of each oil well parameter to the

observation only model and selected the linear or quadratic relationships that performed

better than the observation model for each species.

The top ranked oil well model based on the lowest AICc score was selected for

Brown-headed Cowbird, Chestnut-collared Longspur, Savannah Sparrow, Sprague’s Pipit,

Vesper Sparrow, Western Meadowlark, and Grasshopper Sparrow in 2014 (Table C.1).

Baird’s Sparrow, Bobolink, and Grasshopper Sparrow in 2013 had uninformative

parameters in their top ranked oil well models, so the best competing model was chosen

for these species (Table C.1). No oil well model was selected for Clay-colored Sparrow

and Horned Lark because the observation model was a competing model (Table C.1) and

all oil well variables included zero in their 85% confidence intervals.

166

Table C.1. N-mixture abundance models comparing the most parsimonious oil well models for eleven grassland songbird species in

southeastern SK, 2013-2014. Model variables include: well density (density), well activity (activity), and well proximity (distance).

Oil well models are also compared to the observation (Detection) and Null models. All oil well models include the observation model.

All models are presented with the log likelihood score (LL), number of parameters (K), Akaike’s Information Criterion score adjusted

for small sample sizes (AICc), ∆AICc, and model weight (wi).

Model LL K AICc ∆AICc wi Baird’s Sparrow density + activity + distance adistance + activity distance + density distance activity density + activity density Detection Null

-1122.0 -1123.4 -1127.6 -1130.4 -1135.1 -1135.1 -1144.0 -1148.0 -1248.8

16 15 14 13 14 15 13 12 2

2277.2 2277.8 2284.2 2287.5 2299.2 2301.2 2314.8 2320.7 2501.6

0.0 0.6 6.9

10.3 21.9 24.0 37.6 43.4

224.4

0.56 0.42 0.02 0.00 0.00 0.00 0.00 0.00 0.00

Brown-head Cowbird aactivity Detection Null

-3182.8 -3188.8 -3283.6

13 11 2

6392.4 6400.1 6571.2

0.0 7.7

178.7

0.98 0.02 0.00

Bobolink activity + distance adistance

-1452.4 -1455.3

11 9

2827.4 2929.0

0.0 1.6

0.49 0.22

167

density + activity + distance density + distance activity activity + density density Detection Null

-1452.4 -1455.0 -1463.2 -1463.1 -1467.6 -1469.4 -1507.3

12 10 10 11 9 8 2

2929.4 2930.5 2946.8 2948.9 2953.5 2955.0 3018.5

2.0 3.1

19.4 21.4 26.1 27.6 91.1

0.18 0.11 0.00 0.00 0.00 0.00 0.00

Chestnut-collared Longspur adistance Detection Null

-492.1 -494.9 -572.9

12 11 2

1008.8 1014.4 1149.9

0.0 5.6

141.1

0.94 0.06 0.00

Clay-colored Sparrow activity distance aDetection distance + activity Null

-2300.9 -2302.0 -2303.5 -2300.9 -2455.8

12 11 10 13 2

4626.5 4626.6 4627.4 4628.6 4915.5

0.0 0.0 0.9 2.1

289.0

0.34 0.33 0.22 0.12 0.00

Grasshopper Sparrow 2013 distance2 + density2

adistance2

distance2 + density2 + activity activity + distance2

density2 activity + density2

activity Detection Null

-472.4 -473.6 -472.1 -473.6 -479.5 -477.8 -479.2 -483.1 -624.2

10 9

12 11 9

11 10 8 2

965.7 966.0 969.6 970.3 977.7 978.8 979.3 982.8

1252.5

0.0 0.3 3.9 4.6

12.0 13.1 12.7 17.1

286.8

0.47 0.41 0.07 0.05 0.00 0.00 0.00 0.00 0.00

168

Grasshopper Sparrow 2014 aactivity + distance distance + density + activity distance distance + density activity activity + density density Detection Null

-398.1 -398.0 -405.6 -405.6 -407.7 -407.0 -410.6 -412.1 -482.1

11 12 9

10 10 11 9 8 2

819.4 821.4 830.0 832.2 836.4 837.2 840.0 840.8 968.3

0.0 2.0

10.6 12.8 17.0 17.8 20.6 21.4

148.9

0.73 0.27 0.00 0.00 0.00 0.00 0.00 0.00 0.00

Horned Lark density2 aDetection Null

-577.5 -578.7 -609.3

10 9 2

1175.5 1775.8 1222.6

0.0 0.2

47.0

0.53 0.47 0.00

Savannah Sparrow adistance Detection Null

-2622.2 -2626.2 -2699.2

13 12 2

5271.1 5277.1 5402.5

0.0 6.1

131.4

0.95 0.05 0.00

Sprague’s Pipit adensity2 Detection Null

-583.1 -587.5 -634.0

11 10 2

1188.8 1195.5 1272.0

0.0 6.6

83.2

0.96 0.04 0.00

Vesper Sparrow adistance density + distance Detection density

-1113.3 -1113.3 -1119.8 -1119.2

11 12 10 11

2249.1 2251.2 2260.1 2261.0

0.0 2.1

11.0 11.9

0.74 0.26 0.00 0.00

169

Null

-1162.0 2 2328.0 78.9 0.00

Western Meadowlark adensity2 Detection Null

-1380.9 -1386.9 -1514.1

12 11 2

2786.4 2796.4 3032.3

0.0

10.0 246.0

0.99 0.01 0.00

aMost parsimonious oil well model for each species

170

Table C.2. Parameter estimates of the top oil well models for eleven grassland songbird species in southeastern SK, 2013-2014.

Parameter Estimate Standard Error

Lower 85% CI

Upper 85% CI

Baird’s Sparrow aIntercept adistance activity (abandoned) aactivity (active) activity (no well)

-0.671 0.556

0 0.723 0.206

0.234 0.116

0 0.235 0.206

-1.0080 0.3890

0 0.3846

-0.0906

-0.3340 0.7230

0 1.0614 0.5026

Brown-headed Cowbird aIntercept activity (abandoned) aactivity (active) aactivity (no well)

2.520

0 -0.202 -0.248

0.1153

0 0.0602 0.1111

2.3540

0 -0.2887 -0.4080

2.6860

0 -0.1153 -0.0880

Bobolink Intercept adistance

0.0487 0.2460

0.0592 0.0452

-0.0365 0.1809

0.1339 0.3111

Chestnut-collared Longspur aIntercept adistance

-0.858 0.256

0.1232 0.0917

-1.0354 0.1240

-0.6806 0.3880

Savannah Sparrow aIntercept adistance

1.2750 0.0858

0.0558 0.0298

1.1946 0.0429

1.3554 0.1287

Sprague’s Pipit

171

aIntercept adensity2

-0.800 -0.333

0.124 0.141

-0.9786 -0.5360

-0.6214 -0.1300

Vesper Sparrow Intercept adistance

-0.0375 -0.2322

0.0919 0.0658

-0.1698 -0.3270

0.0948

-0.1374

Western Meadowlark aIntercept adensity2

0.873

-0.093

0.1433 0.0338

0.6666

-0.1417

1.0794

-0.0443

Grasshopper Sparrow 2013 aIntercept adistance2

1.164

-0.318

0.2511 0.0791

0.8024

-0.4319

1.5256

-0.2041

Grasshopper Sparrow 2014 aIntercept activity (abandoned) aactivity (active) activity (no well) adistance

-0.904

0 0.702

-0.635 0.728

0.250

0 0.273 0.462 0.170

-1.2640

0 0.3089

-1.3003 0.4832

-0.5440

0 1.0951 0.0303 0.9728

a85% confidence interval doesn’t include zero

172

Table C.3. Relative importance of each oil well variable in selecting the top oil well

model for eleven grassland songbird species in southeastern SK, 2013-2014.

Oil well parameters Relative Variable Importance Baird’s Sparrow distance activity density

1.00 0.98 0.59

Brown-headed Cowbird activity

0.98

Bobolink distance activity density

1.00 0.69 0.29

Chestnut-collared Longspur distance

0.94

Clay-colored Sparrow activity distance

0.47 0.45

Grasshopper Sparrow 2013 distance density activity

1.00 0.56 0.14

Grasshopper Sparrow 2014 activity distance density

1.00 1.00 0.29

Horned Lark density

0.54

Savannah Sparrow distance

0.96

Sprague’s Pipit density

0.97

Vesper Sparrow distance

1.00

173

density

0.27

Western Meadowlark density

0.99

174

APPENDIX D – SONGBIRD DISTURBANCE MODELS

I used N-mixture abundance models to determine the top disturbance model for

each grassland songbird species. I considered five landscape-scale disturbance

parameters associated with oil development: pipelines, battery and oil building pads, well

pads, oil roads/trails, and cumulative disturbance (sum of all disturbance features). I first

evaluated the interactive and additive effects of year with each disturbance parameter to

determine if I could combine the two years of data together. I then compared the linear

and quadratic effect of each oil well parameter to the observation only model and selected

the linear or quadratic relationships that performed better than the observation model for

each species.

The top ranked disturbance model based on the lowest AICc score was selected

for Baird’s Sparrow, Chestnut-collared Longspur, Clay-colored Sparrow, Grasshopper

Sparrow, Savannah Sparrow, and Western Meadowlark (Table D.1). Bobolink had

uninformative parameters in the top ranked disturbance model, so the best competing

model was chosen (Table D.1). No disturbance parameters were included in models for

Brown-headed Cowbird, Horned Lark, Sprague’s Pipit and Vesper Sparrow because the

observation only model had lower AICc scores than models including disturbance

parameters.

175

Table D.1. N-mixture abundance models comparing the most parsimonious landscape level disturbance models (area disturbed by oil

roads [oil roads], well pads [well pads], battery and oil building pads [bb pads], pipelines [pipelines], total sum of all disturbance

features [cumulative disturbance]), plus the observation only model [Detection] and the Null model in southeastern SK, 2013-2014.

All disturbance models include the observation model. All models are presented with the log likelihood score (LL), number of

parameters (K), Akaike’s Information Criterion score adjusted for small sample sizes (AICc), ∆AICc, and model weight (wi).

Model LL K AICc ∆AICc wi Baird’s Sparrow awell pads Detection Null

-546.9 -550.9 -597.1

13 12 2

1121.4 1127.1 1198.2

0.0 5.8

76.8

0.95 0.05 0.00

Brown-head Cowbird aDetection Null

-1600.2 -1645.2

11 2

3223.5 3294.5

0.0

71.0

1.00 0.00

Bobolink cumulative disturbance + well pads acumulative disturbance well pads Detection Null

-621.0 -622.1 -623.1 -627.6 -657.0

10 9 9 8 2

1262.9 1262.9 1265.0 1271.8 1318.0

0.0 0.0 2.1 9.0

55.1

0.43 0.42 0.15 0.00 0.00

Chestnut-collared Longspur awell pads2 Detection

-183.4 -187.0

12 11

392.1 397.2

0.0 5.0

0.93 0.07

176

Null

-224.0 2 452.0 59.9 0.00

Clay-colored Sparrow acumulative disturbance2 Detection Null

-1125.2 -1128.5 -1209.5

11 10 2

2273.5 2277.9 2423.0

0.0 4.4

149.5

0.90 0.10 0.00

Grasshopper Sparrow abb pads2 + cumulative disturbance2 cumulative disturbance2 bb pads2 Detection Null

-367.8 -371.1 -371.2 -373.6 -460.9

13 12 12 11 2

763.2 767.7 767.7 770.4 925.8

0.0 4.4 4.4 7.2

162.6

0.80 0.09 0.09 0.02 0.00

Horned Lark aDetection Null

-311.7 -320.4

9 2

642.2 644.8

0.0 2.6

0.79 0.21

Savannah Sparrow acumulative disturbance Detection Null

-1278.0 -1282.0 -1316.9

13 12 2

2583.5 2589.3 2637.8

0.0 5.8

54.3

0.95 0.05 0.00

Sprague’s Pipit aDetection Null

-261.7 -285.4

10 2

544.3 574.8

0.0

30.5

1.00 0.00

Vesper Sparrow aDetection Null

-595.6 -628.4

10 2

1212.1 1260.8

0.0

48.7

1.00 0.00

177

Western Meadowlark aoil roads2 Detection Null

-709.1 -711.6 -771.5

12 11 2

1443.6 1446.4 1547.1

0.0 2.8

103.5

0.80 0.20 0.00

aMost parsimonious disturbance model for each species

178

Table D.2. Parameter estimates of the top landscape disturbance N-mixture models of grassland songbirds in southeastern SK, 2013-

2014.

Parameter Estimate Standard Error

Lower 85% CI

Upper 85% CI

Baird’s Sparrow Intercept awell pads

0.0464

-0.3480

0.147 0.134

-0.1653 -0.5410

0.2581

-0.1550

Bobolink aIntercept acumulative disturbance

-0.201 -0.264

0.0892 0.0805

-0.3294 -0.3799

-0.0726 -0.1481

Chestnut-collared Longspur aIntercept awell pads2

-0.611 -0.634

0.248 0.370

-0.9681 -1.1668

-0.2539 -0.1012

Clay-colored Sparrow aIntercept acumulative disturbance2

0.8074 0.0904

0.0792 0.0335

0.6934 0.0422

0.9214 0.1386

Grasshopper Sparrow Intercept abb pads2 acumulative disturbance2

0.0983 0.0347

-0.2355

0.1840 0.0107 0.1021

-0.1667 0.0193

-0.3825

0.3633 0.0501

-0.0885

Savannah Sparrow aIntercept acumulative disturbance

1.252

-0.129

0.0809 0.0455

1.1355

-0.1945

1.3685

-0.0635

179

Western Meadowlark aIntercept aoil roads2

0.716

-0.134

0.1613 0.0658

0.4837

-0.2288

0.9483

-0.0392 a85% confidence interval doesn’t include zero

180

APPENDIX E – AVIAN PREDATOR OBSERVATION MODELS

In developing my avian predator occurrence models, I first selected the most

parsimonious observation model for each avian predator species. I used four detection

parameters in the observation models: ordinal date of survey, time of survey, wind speed

during survey, and observer. There were seven observers over the course of the study,

and one observer performed surveys in 2013 and 2014. The other six observers only

conducted surveys in one of the two study years.

I first compared the interactive and additive effects of year for each detection

parameter to determine if I could combine the two years of data together. I then created a

global model by comparing each detection parameter to the Null model and retaining the

detection parameters that performed better than the Null model. I fitted all subset models

of the global model and selected the top observation model to be used as the detection

component of the occurrance model for each species.

The observation model selection for most species was straightforward. I selected

the top ranked model based on the lowest AICc scores for American Crow, Common

Raven, Northern Harrier, and Red-tailed Hawk (Table D.1). All detection variables in

the selected observation models did not include zero in their 85% confidence intervals

(Table D.2).

I did not select an observation model for Black-billed Magpie or Swainson’s

Hawk. No detection variables performed better than the Null model for Black-billed

Magpie (Table D.1). Swainson’s Hawk had no informative detection variables. No

single detection variable was a good fit in the observation models, and when detection

variables were combined, the fitted models had lower AICc scores than models with

181

single variables (Table D.1). Also, all of the detection variables had low relative

importance for Swainson’s Hawk (Table D.3). Thus, the Null model best explained

detection estimates for Black-billed Magpie and Swainson’s Hawk.

182

Table E.1. N-mixture occurrence models comparing the most parsimonious observation models of three corvid species and three

hawk species in southeastern SK, 2013-2014. Detection covariates include: observer, ordinal date (date), wind speed (wind), and time

of day (time). Models with weights (wi) that sum to 0.90 are shown, plus the Null model. All models are presented with the log

likelihood score (LL), number of parameters (K), Akaike’s Information Criterion score adjusted for small sample sizes (AICc), ∆AICc,

and model weight (wi).

Model LL K AICc ∆AICc wi American Crow aobserver Null

-265.1 -278.0

8 2

546.8 560.0

0.0

13.1

1.00 0.00

Black-billed Magpie Null

-150.9

2

305.8

0.0

1.00

Common Raven aobserver Null

-107.9 -122.8

8 2

232.4 249.7

0.0

17.3

1.00 0.00

Northern Harrier aobserver Null

-142.6 -150.6

8 2

301.8 305.3

0.0 3.6

0.86 0.14

Red-tailed Hawk aobserver + time time observer

-130.2 -140.1 -135.4

9 3 8

279.2 286.3 287.3

0.0 7.1 8.1

0.96 0.03 0.02

183

Null

-146.0 2 296.0 16.8 0.00

Swainson’s Hawk time date wind observer time + date time + wind date + wind date + observer time + observer wind + observer Null

-161.6 -161.6 -161.8 -156.6 -161.4 -161.5 -161.6 -156.4 -156.5 -156.6 -165.3

3 3 3 8 4 4 4 9 9 9 2

329.2 329.3 329.7 329.8 330.9 331.1 331.3 331.5 331.7 332.0 334.6

0.0 0.1 0.4 0.6 1.7 1.9 2.0 2.3 2.5 2.7 5.3

0.16 0.16 0.13 0.12 0.07 0.06 0.06 0.05 0.05 0.04 0.01

aMost parsimonious observation model for each species

184

Table E.2. Parameter estimates of the top observation models for two corvid species and two hawk species in southeastern SK, 2013-

2014.

Parameter Estimate Standard Error

Lower 85% CI

Upper 85% CI

American Crow aIntercept obs1 obs2 obs3 obs4 obs5 aobs6 obs7

-1.1961

0 0.3458 0.6194 0.1539 0.6306 1.8987 0.0165

0.357

0 0.491 0.614 0.445 0.614 0.485 0.600

-1.7102

0 -0.3612 -0.2648 -0.4869 -0.2536 1.2003

-0.8475

-0.6820

0 1.0528 1.5036 0.7947 1.5148 2.5971 0.8805

Common Raven aIntercept obs1 obs2 aobs3 aobs4 obs5 obs6 obs7

-3.064

0 0.746 1.882 1.216

-0.676 -11.174 -10.256

0.974

0 0.897 0.836 0.794 1.255

187.319 120.097

-4.4666

0 -0.5457 0.6782 0.0726

-2.4832 -280.9134 -183.1957

-1.6614

0 2.0377 3.0858 2.3594 1.1312

258.5654 162.6837

Northern Harrier aIntercept obs1 obs2

-3.3782

0 1.2728

1.07

0 1.19

-4.9190

0 -0.4408

-1.8374

0 2.9864

185

aobs3 aobs4 aobs5 aobs6 obs7

2.3299 2.2295 1.7830 2.3676 0.0658

1.14 1.07 1.15 1.11 1.45

0.6883 0.6887 0.1270 0.7692

-2.0222

3.9715 3.7703 3.4390 3.9660 2.1538

Red-tailed Hawk aIntercept obs1 aobs2 obs3 aobs4 aobs5 obs6 obs7 atime

-2.947

0 3.266 1.536 1.716 1.977 1.531

-0.228 0.776

1.070

0 1.180 1.282 1.134 1.271 1.176 1.526 0.297

-4.4878

0 1.5668

-0.3101 0.0830 0.1468

-0.1624 -2.4254 0.3483

-1.4062

0 4.9652 3.3821 3.3490 3.8072 3.2244 1.9694 1.2037

a85% confidence interval doesn’t include zero

186

Table E.3. Relative importance of each detection variable in selecting the top observation

model for two corvid species and three hawk species in southeastern SK, 2013-2014.

Detection Parameters Relative Variable Importance American Crow observer

1.00

Common Raven observer

1.00

Northern Harrier observer

1.00

Red-tailed Hawk time observer

0.98 0.98

Swainson’s Hawk time date observer wind

0.41 0.41 0.39 0.36