Citation
Patterns of occurrence and co-occurrence for swift fox (vulpes velox), western burrowing owl (athene cunicularia hypugaea), and mountain plover (charadrius montanus) on black-tailed prairie dog (cynomys ludovicianus) colonies : a trend data summary and a hierarchical occupancy analysis

Material Information

Title:
Patterns of occurrence and co-occurrence for swift fox (vulpes velox), western burrowing owl (athene cunicularia hypugaea), and mountain plover (charadrius montanus) on black-tailed prairie dog (cynomys ludovicianus) colonies : a trend data summary and a hierarchical occupancy analysis
Creator:
Parker, Ryan Andrew
Place of Publication:
Denver, CO
Publisher:
University of Colorado Denver
Publication Date:
Language:
English

Thesis/Dissertation Information

Degree:
Master's ( Master of science)
Degree Grantor:
University of Colorado Denver
Degree Divisions:
Department of Integrative Biology, CU Denver
Degree Disciplines:
Biology
Committee Chair:
Wunder, Michael B.
Committee Members:
Tomback, Diana F.
Hartley, Laurel M.
Dwyer, Angela M.

Notes

Abstract:
Swift fox (Vulpes velox), Burrowing Owl (Athene cunicularia) and Mountain Plover (Charadrius montanus) are three species of conservation concern who all co-occur on blacktailed prairie dog (Cynomys ludovicianus) colonies across the Great Plains of North America. In this thesis, we examine the relationships between these species by 1) proposing an investigation into food-web interactions between foxes, owls, and plovers found specifically on prairie dog colonies, and 2) examining influences on occurrence of foxes, owls, and plovers on prairie dog colonies. Literature to-date heavily focuses on each of these species individually, but a knowledge gap exists when considering multi-species ecology in this system. Exploring how multiple co-occurring species interact in Great Plains systems will provide a better understanding of the ecology across this fragile landscape. Chapter One explores the possibility of foxes, owls, and plovers interacting across trophic-levels. Trophic cascades occur when flora and fauna directly and/or indirectly influence co-occurring species populations at different levels of the food chain, and North American temperate grasslands provide an interesting case study to research these relationships. We briefly define trophic cascades in terrestrial systems and explore the potential for a cascading trophic interaction between grassland associated swift fox, burrowing owl, and mountain plover, three rangeland species of conservation concern, on black-tailed prairie dog colonies using two U.S. Forest Service data sets. Historic patterns of occurrence and co-occurrence suggest top-down control governs the spatiotemporal distribution patterns of the three species and may be influenced by habitat fragmentation and management actions. Managing for interactive, multi-trophic communities requires the identification of species interactions and the mechanisms that drive them. Long-term multi-species occupancy research combined with hypothesized driving mechanisms and the co-occurrence of associated grassland species is recommended for addressing these complex interactions moving forward. Chapter Two focuses on the occurrence of foxes, owls, and plovers on prairie dog colonies. Co-occurring species often benefit from the same patches of habitat, and in some cases may be involved in complex-interactions driven by patch-characteristics. These characteristics can include but are not limited to patch-size and resource-availability, which can be coupled with the presence or absence of multiple co-occurring species to evaluate probability estimates of occupancy. Prairie dog colonies serve as patches of habitat and swift fox, burrowing owl, and mountain plover co-occur on these colonies benefiting from the shelter and food resources provided. In this chapter, we derive probability estimates for occupancy of prairie dog colonies by foxes, owls, and plovers in eastern Wyoming. We hypothesize that co-occurring species presence or absence of two species on a colony influences the occupancy of the third species, and that colony size also drives occurrence. We use hierarchical modeling to develop single-species occupancy models for each species across n=88 prairie dog colonies pooled across three sampling seasons (2016-2018). We were unable to identify effects on probability estimates of occupancy for foxes, likely because sparse data prevented model convergence. For owls and plovers, we identified an effect of colony size on occupancy. For every 1-unit ha increase in colony size, the odds that a colony was occupied by an owl and plover increased by 1.403 ± 1.358 and 7.637 ± 2.854, respectively. Our sparse data set likely prevented proper model convergence for the remaining owl and plover models. Our results highlight that occupancy-driven analyses require more-robust data sets spanning multiple years of sampling and an increased number of sample units. We cannot confirm nor deny that foxes, owls, and plovers are involved in complex-interactions within a patchy-environment, and our results warrant a long-term, multi-species experiment in this system.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
Copyright Ryan Andrew Parker. Permission granted to University of Colorado Denver to digitize and display this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.

Downloads

This item has the following downloads:


Full Text
PATTERNS OF OCCURRENCE AND CO-OCCURRENCE FOR SWIFT FOX (VULPES VELOX), WESTERN BURROWING OWL (ATHENE CUNICULARIA HYPUGAEA), AND MOUNTAIN PLOVER (CHARADRIUS MONTANUS) ON BLACK-TAILED PRAIRIE DOG (CYNOMYS LUDOVICIANUS) COLONIES: A TREND DATA SUMMARY AND A HIERARCHICAL OCCUPANCY ANALYSIS
By
RYAN ANDREW PARKER B.S., University of Wyoming, 2015
A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Master of Science Biology Program
2019


This thesis for the Master of Science degree by Ryan Andrew Parker has been approved for the Biology Program by
Michael B. Wunder, Chair Diana F. Tomback Laurel M. Hartley Angela M. Dwyer


Parker, Ryan Andrew (M.S., Biology Program)
Patterns of Occurrence and Co-Occurrence for Swift Fox (Vulpes velox), Western Burrowing Owl {Athene cunicularia hypugaea), and Mountain Plover {Charadrius montanus) on Blacktailed Prairie Dog (Cynomys ludovicianus) Colonies: A Trend Data Summary and a Hierarchical Occupancy Analysis
Thesis directed by Associate Professor Michael B. Wunder
ABSTRACT
Swift fox {Vulpes velox), Burrowing Owl {Athene cunicularia) and Mountain Plover {Charadrius montanus) are three species of conservation concern who all co-occur on blacktailed prairie dog {Cynomys ludovicianus) colonies across the Great Plains of North America. In this thesis, we examine the relationships between these species by 1) proposing an investigation into food-web interactions between foxes, owls, and plovers found specifically on prairie dog colonies, and 2) examining influences on occurrence of foxes, owls, and plovers on prairie dog colonies. Literature to-date heavily focuses on each of these species individually, but a knowledge gap exists when considering multi-species ecology in this system. Exploring how multiple co-occurring species interact in Great Plains systems will provide a better understanding of the ecology across this fragile landscape.
Chapter One explores the possibility of foxes, owls, and plovers interacting across trophic-levels. Trophic cascades occur when flora and fauna directly and/or indirectly influence co-occurring species populations at different levels of the food chain, and North American temperate grasslands provide an interesting case study to research these relationships. We briefly
iii


define trophic cascades in terrestrial systems and explore the potential for a cascading trophic interaction between grassland associated swift fox, burrowing owl, and mountain plover, three rangeland species of conservation concern, on black-tailed prairie dog colonies using two U.S. Forest Service data sets. Historic patterns of occurrence and co-occurrence suggest top-down control governs the spatiotemporal distribution patterns of the three species and may be influenced by habitat fragmentation and management actions. Managing for interactive, multi-trophic communities requires the identification of species interactions and the mechanisms that drive them. Long-term multi-species occupancy research combined with hypothesized driving mechanisms and the co-occurrence of associated grassland species is recommended for addressing these complex interactions moving forward.
Chapter Two focuses on the occurrence of foxes, owls, and plovers on prairie dog colonies. Co-occurring species often benefit from the same patches of habitat, and in some cases may be involved in complex-interactions driven by patch-characteristics. These characteristics can include but are not limited to patch-size and resource-availability, which can be coupled with the presence or absence of multiple co-occurring species to evaluate probability estimates of occupancy. Prairie dog colonies serve as patches of habitat and swift fox, burrowing owl, and mountain plover co-occur on these colonies benefiting from the shelter and food resources provided. In this chapter, we derive probability estimates for occupancy of prairie dog colonies by foxes, owls, and plovers in eastern Wyoming. We hypothesize that co-occurring species presence or absence of two species on a colony influences the occupancy of the third species, and that colony size also drives occurrence. We use hierarchical modeling to develop singlespecies occupancy models for each species across n=88 prairie dog colonies pooled across three sampling seasons (2016-2018). We were unable to identify effects on probability estimates of
IV


occupancy for foxes, likely because sparse data prevented model convergence. For owls and plovers, we identified an effect of colony size on occupancy. For every 1-unit ha increase in colony size, the odds that a colony was occupied by an owl and plover increased by 1.403 ±
1.358 and 7.637 ± 2.854, respectively. Our sparse data set likely prevented proper model convergence for the remaining owl and plover models. Our results highlight that occupancy-driven analyses require more-robust data sets spanning multiple years of sampling and an increased number of sample units. We cannot confirm nor deny that foxes, owls, and plovers are involved in complex-interactions within a patchy-environment, and our results warrant a longterm, multi-species experiment in this system.
The form and content of this abstract are approved. I recommend its publication.
Approved: Michael B. Wunder
v


ACKNOWLEDGEMENTS
A huge thank you goes out to my advisor, Dr. Michael B. Wunder, for his guidance on scientific literacy and writing, experimental design, ecological data analysis, career building, critical and independent thinking, and time management. A huge thank you to Dr. Diana F. Tomback and Dr. Laurel M. Hartley for their guidance and hours of discussion on my system of study, scientific methods, navigating a M.S. thesis, and scientific writing and communication. Another huge thank you to Angela M. Dwyer with Bird Conservancy of the Rockies for guidance on my system of study, scientific writing and communication, funding for field work, and for seeing the importance of my graduate work as it applies to the conservation and ecology of the North American Great Plains. I would also like to give a huge thank you to Cristi Painter with the U.S. Forest Service for being instrumental in developing, supporting, and seeing this graduate project through till the end on USFS lands in eastern Wyoming, and also for being a mentor and amazing friend. To my fellow graduate student friends, thank you for countless hours of discussion, critical thinking, constructive advice, and mental support throughout the course of my project. In particular, I would like to thank Ben Lagasse, Sara St. Onge, Tyler Michels, Alii Pierce, Scott Yanco, Amber Carver, Libby Pansing, Michelle Deprenger-Levin, Andrew McDevitt, Katie Kilpatrick, David Schutt, Marianne Davenport, Mac Calvert, and Katherine Fu. To Timothy Fosmo and Joseph Alfonso, a huge thank you for helping collect data in the field for my project. An enormous thank you goes out to my friends and family who provided me with much needed mental and emotional support for the duration of my mater’s program. And finally, my biggest thank you goes out to my partner and number one fan, Christopher Gilbert.
All wildlife research methods were approved by the University of Colorado Institutional Animal Care and Use Committee (IACUC Protocol # 92014(05)1C)
vi


TABLE OF CONTENTS
CHAPTER
I. TROPHIC ECOLOGY WARRANTS MULTI-SPECIES MANAGEMENT IN A
GRASSLAND SETTING: PROPOSED SPECIES INTERACTIONS ON BLACK-TAILED PRAIRIE DOG COLONIES..................................1
Introduction.................................................1
North American Temperate Grasslands..........................2
Focal Species................................................4
Understanding Trophic Ecology................................6
CASE STUDY...................................................9
Trophic Interactions on the Pawnee and Thunder Basin National
Grasslands...................................................9
Hypotheses Concerning Trophic Ecology on Pawnee and Thunder Basin
National Grasslands.........................................12
Additional Influences from USFS Grassland Management Techniques...20 Conclusions.................................................21
II. FACTORS INFLUENCING THE OCCURRENCE OF SWIFT FOX,
BURROWING OWL, AND MOUNTAIN PLOVER ON BLACK-TAILED PRAIRIE DOG COLONIES IN WYOMING...................................22
Introduction................................................22
Methods.....................................................26
Results.....................................................33
Discussion..................................................36
vii


REFERENCES........................................................41
APPENDIX..........................................................48
viii


CHAPTERI
TROPHIC ECOLOGY WARRANTS MULTI-SPECIES MANAGEMENT IN A GRASSLAND SETTING: PROPOSED SPECIES INTERACTIONS ON BLACKTAILED PRAIRIE DOG COLONIES
Introduction
North American temperate grasslands support high levels of species richness and diversity and are an important resource for agricultural production, so research and appropriate management for these rangeland systems necessarily must focus on the interactions between species, rather than on single-species populations (Samson and Knopf 1994, Augustine and Baker 2013, Grant et al. 2017). Because species-interaction paradigms are crucial for management in these systems, we present a theoretical framework to address a broad issue: that cascading trophic interactions play a critical role in structuring populations of concern in grasslands. As a case study, we examine the hypothesis that co-occurring swift fox (Vulpes velox), western burrowing owl (Athene cunicularia hypugaea) and mountain plover (Charadrius montanus) are involved in a cascading trophic interaction on black-tailed prairie dog (Cynomys ludovicianus) colonies on North American temperate grasslands. Trophic cascades among these species remain undocumented and are worthy of investigation for effective multi-species management in these communities of conservation concern. We begin by introducing North American temperate grasslands and follow with short species accounts for prairie dogs, foxes, owls, and plovers. We then briefly define trophic interactions, proceeded by a case study on the Pawnee National Grassland (PNG) in Colorado and Thunder Basin National Grassland (TBNG) in Wyoming. Finally, we hypothesize potentially influencing mechanisms in a fox-owl-plover
1


interaction. We fully acknowledge that inferences based on the trend data and literature cited in this paper are anecdotal and statistically insufficient and maintain that future investigation into this system is needed. Collectively, this paper serves as a proposal for future trophic-level research in this interactive community.
North American Temperate Grasslands
North American temperate grasslands historically covered 162 million hectares (ha) of central and western North America (Samson and Knopf 1994). These grasslands can be loosely subdivided based on characteristics of the dominant plant life forms into shortgrass, mixed-grass, or tail-grass prairies (Coupland 1961, Samson and Knopf 1994, Lauenroth and Burke 2008; Fig. 1). In this review we focus our attention on the shortgrass and northern mixed-grass prairies of the Great Plains. The shortgrass prairie, or shortgrass steppe ecoregion, is dominated by C4 grasses (typically 6-12 ” in height) including blue grama (Bontelona gracilis), buffalo grass (Bouteloua dactyloides), and Western wheat grass (Pascopyrum smithii) and comprises the southwestern portion of North American temperate grasslands (Lauenroth et al. 1999, Lauenroth and Burke 2008; Fig. 1). Mixed-grass prairies are split into two sub-regions: northern mixed-grass and southern mixed-grass, extending north into Canada and south to northcentral Texas (Lauenroth et al. 1999, Lauenroth and Burke 2008; Fig. 1). Mixed-grass prairies are characterized by ecotones of shortgrass steppe and sagebrush shrubland, and support rich communities of forbs like goldenrod (Solidago sp.), slimflower scurfpea (Psoralea tenuiflora), and scarlet globemallow (Sphaeralcea coccinea) (Lauenroth et al. 1999, Adler and HilleRisLambers 2008). Common shrub species may include Wyoming big sagebrush (Artemisia tridentata wyomingensis) and greasewood (Sarcobatus vermiculatus) (Lauenroth et al. 1999,
2


Adler and HilleRisLambers 2008). Collectively, the vegetation structure of these grasslands support diverse communities of interactive fauna (Samson and Knopf 1994, Augustine and Baker 2013, Grant et al. 2017), presenting us with an opportunity to highlight potential trophic-level cascades that inform future grassland management objectives.
Figure 1. Map of North American temperate grassland categories and regions. Pawnee National Grassland represents a shortgrass prairie and Thunder Basin National Grassland represents a northern mixed-grass prairie. Data retrieved from Esri REST Services Directory.
3


Focal Species
Our hypothesis concerning trophic cascades in this system hinges on the role of blacktailed prairie dogs in structuring requisite habitat for these species. The ecological benefits of black-tailed prairie dogs are extensive: this species engineers resource-rich environments for grassland birds, small to medium sized carnivores, and many burrowing species by providing habitat for breeding, rearing of offspring, foraging, and cover from predators (Plumb et al. 2005, Hoogland 2006, Martin et al. 2007, Thiele et al. 2013). The US Fish and Wildlife Service (USFWS) reported a historically-low 364,000 active hectares of prairie dog colonies in 1961 (USFWS 2002, Virchow and Hygnstrom 2002), and more recent estimates suggest a range-wide decline of 90-98% (Augustine and Baker 2013), leading to their status of species of conservation concern (CPW 2015, USDA 2017, WGFD 2017). These declines have also been linked to decreasing populations of swift fox, burrowing owl, and mountain plover, three grassland species of conservation concern with the potential for trophic interactions (Desmond et al. 2000, Nicholson et al. 2006, Augustine et al. 2008, Alverson and Dinsmore 2014).
Swift Fox
The swift fox (Vulpes velox; Fig. 2) is a small predatory canid that frequently occupies prairie dog colonies and relies on denning opportunities and abundant prey resources therein (Kitchen et al. 1999, Kintigh and Anderson 2005). Shrub-free expanses across the shortgrass prairie like open fields and burned areas provide additional habitat for swift fox (Thompson et al. 2008). In regions characterized by mixed-grass prairie and sagebrush steppe, swift fox may be specifically tied to prairie dog colonies because shrub composition and structure in these
4


landscapes is low. Swift fox are a species of conservation concern at both state and federal levels (CPW 2015, USD A 2017, WGFD 2017).
Burrowing Owl
Western burrowing owls (Athene cunicularia hypugaea; Fig. 2) are small, fossorial raptors that lay eggs, rear chicks, and shelter in pre-excavated burrow systems (Poulin et al. 2011). Today, burrowing owls regularly colonize urban environments (e.g., golf courses, landfills, and storm drainage systems), remain abundant in agroecosystems, and frequently associate with both urban and rural prairie dog complexes (Smith et al. 2005, Williford et al. 2009, Manning 2011, Maclas-Duarte and Conway 2015, Chi 2016). In the Great Plains, owls have been documented to respond to prairie dog alarm calls, providing evolutionary evidence that owls use prairie dogs for detecting predators (Bryan and Wunder 2014). Prairie dog colonies also support populations of ground nesting birds and insects; both prey items for burrowing owls. Burrowing owls remain a species of conservation concern at state and federal levels (CPW 2015, USDA 2017, WGFD 2017), and North American Breeding Bird Survey (BBS) data between 1966-2011 estimate an annual population decline of 1.1% (Sauer et al. 2013).
Mountain Plover
Mountain plover (Charadrius montanus; Fig. 2) are upland shorebirds that nest, rear young, and forage on landscapes with a relatively high bare-ground structure, taking advantage of a prairie dog’s natural tendency to clip vegetation and create short-grass habitat (Augustine and Baker 2013). Plovers do not exclusively depend on prairie dog colonies for breeding habitat, as pastures moderately to heavily grazed by cattle, burned areas, and fallow crop fields throughout the southern portion of the species range provide favored habitat requirements
5


(Wunder et al. 2003, Augustine and Skagen 2014, Ramsdell et al. 2016). Prairie dog colonies; however, provide a greater source of habitat for breeding mountain plovers in the mixed-grass prairies of the Northern Great Plains (Wyoming, Montana, Canada; Dinsmore et al. 2005, Plumb et al. 2005). Here, shortgrass prairies transition into mixed-grass/sagebrush steppe landscapes where prairie dogs often create the only shortgrass/bare-ground habitat suitable for nesting. Mountain plovers remain a species of conservation concern throughout their range (CPW 2015, USDA 2017, WGFD 2017) and North American Breeding Bird Survey (BBS) data between 1966-2011 estimate an annual population decline of 3.0% (Sauer et al. 2013).
Figure 2. The swift fox (Viilpes velox\ left), western burrowing owl {Athene cumcularia hypugaea; center) and mountain plover {Charadrius montanus; right) all directly benefit from black-tailed prairie dogs {Cynomys ludovicianus) where their respective ranges overlap (Image credit: Cristi Painter).
Understanding Trophic Ecology
Management objectives for prairie dogs should be geared towards meeting the requirements of multiple associated species of conservation concern. Reaching these objectives first requires a careful understanding of potential cascading trophic interactions between species.
6


Trophic Cascades
Trophic cascades can be classified as facilitating top-down or bottom-up responses (Elmhagen and Rushton 2007). Top-down control is driven by apex or dominant predators (predators at the top of the food chain with no direct threat of predation) or quaternary consumers (species at the top-level of the food chain capable of consuming all lower-level species; Lindeman 1942) triggering a response from lower-level predators, consumers, or producers in the food chain (Elmhagen and Rushton 2007, Ripple et al. 2016; Fig. 3). Bottom-up control is characterized by primary producers or lower-level consumers driving changes to patterns in plant or wildlife populations and distributions higher up the food chain (Elmhagen and Rushton 2007, Ripple et al. 2016; Fig. 3). Here, we focus on top-down trophic cascades facilitating responses in plant or wildlife communities that may in turn structure food chains and ecosystems (Paine 1980, Elmhagen and Rushton 2007, Ripple et al. 2016). We do not discount; however, that bottom-up mechanisms may be simultaneously at work and we support this by evaluating impacts to the vegetation community that may cause upward influences along the food-chain. Responses may manifest as oscillations over time, as constant increases or decreases, or as static in populations across all trophic levels (Polis and Holt 1992, Jiang et al. 2009). These responses have repercussions for management, particularly when communities are comprised of multiple species of conservation concern (Soule et al. 2005).
7


Top-Down Interactions
/ Upper Level Apex \ f Predator/Quaternary J
\, Consumer Controls: J
Lower Level
Predator/Consumer/ ) Producer
r\
imer/ 1
r Upper Level \
Predator/Consumer )
Bottom-Up Interactions
Figure 3. General processes of top-down interactions {left) where upper trophic-level species control populations of lower trophic-level species, and bottom-up interactions {right) where lower trophic-level species control populations of upper trophic-level species (Elmhagen and Rushton 2007, Ripple et al. 2016).
Intraguild Predation and Mesopredator Release
Top-down control occurs in many predator/prey relationships where intraguild predation and mesopredator release is present. We define intraguild predation as predators across different trophic-levels competing for a shared prey resource, while the lower trophic-level predators also risk predation from the upper trophic-level predators (Polis and Holt 1992). This process fits within a broader, widely-accepted predator/prey hypothesis commonly referred to as mesopredator release. Here, apex predators become rare or absent in an ecosystem resulting in population spikes of smaller, medium-level (meso) predators (Crooks and Soule 1999, Prugh et al. 2009). Mesopredators exploit prey resources at little cost of predation or competition from apex predators and initiate top-down control on lower trophic-level prey communities (Crooks and Soule 1999, Elmhagen and Rushton 2007, Prugh et al. 2009, Ripple et al. 2016; Fig. 3). To
8


further explore these hypotheses, we introduce a case study focusing on the dynamics between swift fox, burrowing owl, and mountain plover in the grasslands of the Great Plains.
Case Study
Trophic Interactions on the Pawnee and Thunder Basin National Grasslands
To examine potential cascading trophic interactions within grassland systems we explored raw-data trends (Fig. 4) in annual prairie dog colony mapping (i.e. total colony boundary; surface area measured in hectares (ha)) and count data for swift fox, burrowing owl, and mountain plover for two federally managed U.S. Forest Service (USFS) National Grasslands; (1) the Pawnee National Grassland (PNG) in Weld County of northeastern Colorado representing a shortgrass prairie (USFS 2016a; Fig. 1), and (2) the Thunder Basin National Grassland (TBNG) in Converse, Weston, and Campbell Counties of eastern Wyoming representing a northern mixed-grass prairie (USFS 2016b; Fig. 1).
Data Collection
Both data sets are used annually by the USFS to inform management of viable populations of prairie dogs and associated species (USFS 2016a, USFS 2016b), and are collected under a collaborative effort by university researchers, federal wildlife managers, and trained volunteers. Total prairie dog colony hectares (ha) on both USFS grasslands are attained by driving the outer boundary of active colonies and mapping those boundaries with GPS (Global Positioning System) units between June and September. Active colonies are determined by the presence of prairie dogs, to include: fresh scat, freshly clipped vegetation, fresh digging at burrow entrances, and visual/aural identification. Only burrows that indicate evidence of active
9


prairie dogs are included in the mapping effort. Mountain plover and burrowing owl surveys occur in early morning and late evening by driving transects no more than 400m apart across prairie dog colonies, stopping every 30-60 seconds to scan for the presence of adult birds during the breeding season (mid-May through mid-July). Swift fox surveys are conducted mid-August through September via spotlighting at night while driving transects across prairie dog colonies or along designated USFS grassland roads adjacent to prairie dog colonies to detect and confirm presence by eye shine. Surveys are conducted one time per colony, and total observations per colony are included into the annual data set. All survey methods follow USFS survey protocol (USFS 2016a, USFS 2016b). Count data belongs to the US Forest Service and will be made available upon request.
2004 2007 2010 2013 2016
Year
1998 2001
2004 2007 2010 2013
Year
_14000 £12000 $10000 « 8000
>, 4000 | 2000 S 0
Prairie Dogs 300 Fox j
| 200 — — Owl /
Plover /
£ 150 O 100
% 50 0
2010 2011 2012 2013 2014 2015 2016 Year
2010 2011 2012 2013 2014 2015 2016 Year
Figure 4. Annual count data in raw-form collected on (a.) the Pawnee National Grassland (PNG) from 1998-2016 and (b.) the Thunder Basin National grassland (TBNG) from 2010-2016 for total active prairie dog colonies measured in hectares (ha) and population counts for swift fox, burrowing owl, and mountain plover (USFS 2016a, USFS 2016b).
10


These raw data (Fig. 4) were collected under protocols targeted for agency-management purposes, and not designed to address research objectives (USFS 2016a, USFS 2016b); therefore, inferences concerning processes in trophic ecology are purely speculative. We fit an exponential curve to the data (Fig. 5) to heuristically observe patterns suggesting that the total area of active prairie dog colonies has increased across the PNG, with number of observed burrowing owls increasing and swift foxes and mountain plovers remaining uncommon and moderately declining (1998-2016; Fig. 5a). Total area of active prairie dog colonies has also increased across TBNG; however, number of observed swift fox and mountain plover are increasing and burrowing owls are only responding weakly (2010-2016; Fig. 5b). Combining these observed co-occurrence patterns with examples of trophic ecology defined in the literature suggest the possibility of a cascading trophic interaction in this system, where the presence or absence of swift fox directly influences the abundance of mountain plovers by controlling burrowing owl mesopredators. Factors beyond increased prairie dog colony area are likely at play because species responses differ markedly between the PNG and TBNG during the same period.
11


a. PNG
300
Fox
2000
£
^ 1500
•
i.
IB
0 1000 ®
£
§ 500
o
o
0
1998 2001 2004
2007
Year
2010 2013
200 100 0
1300 > 200
5 100 o
* 0
300
200
2016 100 0
Plover
1998 2001 2004 2007 2010 2013 2010 Year
b. TBNG
18000 16000 ~14000 • 12000 " 10000 8000 6000 4000 2000 0
2010
o
0)
£
>*
|
0
0
o
2011 2012
2013
Year
2014 2015 2016
2010 2012 2014 2016
Year
Figure 5. An exponential curve fitted to annual trend data collected on (a.) the Pawnee National Grassland (PNG) from 1998-2016 and (b.) the Thunder Basin National grassland (TBNG) from 2010-2016 for total active prairie dog colonies measured in hectares (ha) and population counts for swift fox, burrowing owl, and mountain plover (USFS 2016a, USFS 2016b).
Hypotheses Concerning Trophic Ecology on Pawnee and Thunder Basin National Grasslands
Mesopredator Release on the PNG
Mesopredator release may explain the increase of burrowing owls on the PNG and their co-occurrence patterns with swift fox and mountain plover. Burrowing owls may respond positively to declining swift fox populations (Fig. 5a); ‘releasing’ them from the predation pressure of a higher-level predator. Hunting and trapping presents an annual additive mortality
12


factor for swift fox in Colorado with an estimated harvest of over 600 individuals in the 2014-
2015 season (CPW 2016, Stukel 2017); however, uncertainty around how to obtain accurate harvest estimates of swift fox remains a challenge for state wildlife agencies (CPW 2016). The current population of swift fox in eastern Colorado is unclear; however, state-managed occupancy surveys occurring every 5 years suggests that eastern Colorado still supports the largest population of swift fox throughout their range (Stratman 2017). Predation from coyotes is also a driver of swift fox mortality in Eastern Colorado (Kitchen et al. 1999, Karki et al. 2007). Kitchen et al. found that coyotes contributed to 48% of swift fox mortality in a resource partitioning study in southeastern Colorado (Kitchen et al. 1999). Complex trophic interactions between canid sp. are documented in the literature (Robinson et al. 2014, Lonsinger et al. 2017); however, the extent to which coyotes currently impact swift fox on the PNG remains uncertain. Because coyotes have been documented to influence swift fox abundance (Kitchen et al. 1999, Karki et al. 2007), coyote presence should be considered as a driving mechanism in future trophic research between foxes, owls, and plovers (Fig. 6) on black-tailed prairie dog colonies on the PNG.
13


Figure 6. (a.) Swift Fox (Viilpes velox), (b.) Western Burrowing Owl {Athene cumcularia hypngaea), and (c.) Mountain Plover (Charadrius montanus) are all associated species of (d.) Black-tailed Prairie Dog {Cynomys hidoviciamis) colonies. We hypothesize that foxes consume owls (solid arrow) and owls consume plovers (solid arrow). Foxes also consume plovers (dashed arrow) and prairie dogs (dotted arrow), making this a highly interactive predator/prey community, (e.) Coyotes {Canis latrans) are also present in the system and play an unknown role in the hypothesized cascade by potentially influencing swift fox abundance, and preying opportunistically on all species (Image credit: Cristi Painter, Ryan A. Parker, and © Can Stock Photo Inc.).
Regardless of the cause, a decline in the count of swift fox on the PNG is concurrent with the increased count of burrowing owls and decreased count of mountain plover (USFS 2016a; Fig. 5a). Release of owl mesopredators may be linked to increasing infrequency of mountain plovers on prairie dog colonies on the PNG despite expansion of suitable habitat. Lower trophic-level prey resources often decline after the eruption of a mesopredator due to increased predation (Crooks and Soule 1999, Prugh et al. 2009). Burrowing owls consume mountain plover nestlings (Conrey 2010) supporting this hypothesized trophic cascade between foxes, owls, and plovers (Fig. 6). We also note that radio-transmitters attached to plovers have been recovered inside swift
14


fox dens, suggesting that plovers are a shared prey-resource for both foxes and owls (Miller and Knopf 1993, Conrey 2010). Under this scenario, an increased predation effect on plovers by owls may outcompete a predation effect on plovers by foxes.
Mesopredator Release on the TBNG
Predator/prey dynamics consistent with intraguild predation (Polis and Holt 1992) compare with co-occurrence patterns seen on TBNG. Swift fox remain common and protected from hunting and trapping in the northern mixed-grass prairies of eastern Wyoming (USFS 2016b, WGFD 2017), and spotlight surveys suggest that this region remains a stronghold for swift fox. Although burrowing owl presence indicates a moderate increase on TBNG, trends remain low despite increasing availability of suitable habitat (USFS 2016b; Fig. 5b). Mountain plover populations mirror the trend of swift fox, suggesting that even a minor increase in the count of burrowing owls still potentially alleviates predation on plovers which allows them to take advantage of expanding prairie dog colonies, reflecting patterns of intraguild predation between owls and foxes (USFS 2016b; Fig. 5b). Anecdotally, burrowing owls are selecting for resource deficient habitats where swift fox are less likely to persist; the shared prey resource, mountain plover, is responding positively to the owls change in habitat selection and the increased acreage of prairie dog colonies.
Coyotes are also present on the landscape on TBNG; however, the degree of impact on
lower trophic-level canid species remains undocumented. Coyote populations are lethally controlled in eastern Wyoming by local landowners, hunters, and predator control groups and are infrequently documented on prairie dog colonies during associated species abundance surveys
15


(USFS 2016b). Successfully controlled coyote populations on the TBNG provides support for an increased abundance of swift fox; however, this inference remains speculative based on conversations with USFS biologists (USFS 2016b). Because coyotes have been documented to influence swift fox abundance in Colorado (Kitchen et al. 1999, Karki et al. 2007), coyote presence should be considered as a driving mechanism in future trophic research between foxes, owls, and plovers (Fig. 6) on the TBNG in Wyoming.
Landscape Fragmentation Driving Mesopredator Release
Conversion of grassland habitats may be an important driver of predator/prey dynamics because cascading trophic interactions have been linked to landscape fragmentation (Crooks and Soule 1999, Prugh et al. 2009). Prugh et al. (2009) point to three factors that lead to declines in populations of top-level predators in fragmented landscapes: (1) Apex predators have expansive territories and need large, connected areas of in-tact habitat (Prugh et al. 2009). Therefore, as landscapes become fragmented, apex predators are less likely to meet resource requirements and become uncommon or absent from the area. (2) Fragmented landscapes provide attractive habitat for mesopredators because apex predators become rare and food resources in the form of human waste or crop fields become plentiful (Prugh et al. 2009). And (3), in fragmented landscapes, apex predators experience frequent human conflict leading to lethal predator control (Prugh et al. 2009). Thus, we uphold that predator-driven dynamics and the degree of habitat fragmentation could also support our hypothesis that swift fox, burrowing owl, and mountain plover are involved in a complex trophic interaction on black-tailed prairie dog colonies (Fig. 6).
Landscape Fragmentation on the PNG
16


Shortgrass prairie on the PNG is intermixed with independently managed rangeland and privately-owned agricultural cropland, the latter fragments the landscape by transforming native grassland into cropland. Here, individual prairie dog colonies remain small (-0.5-138 ha, 2016 mapping data) and dispersed (USFS 2016a). Spatial and temporal changes to prairie dog colonies partly result from agricultural development, but also from Yersiniapestis, an infectious bacterium that causes sylvatic plague (a disease often lethal to prairie dogs; Eads and Biggins 2017). Plague is present annually on the PNG, clearing out over 95% of the prairie dogs in any given colony (Webb et al. 2006, Eads and Biggins 2017). Annual prairie dog mapping reveals periodic appearance and disappearance of individual active colonies that shift geographically across the landscape between years (USFS 2016a). Thus, a combination of agricultural development and plague contributes to the fragmentation and small size of prairie dog colonies.
The total number of small, isolated colonies are rising on the PNG (USFS 2016a); however, suggesting that benefits to swift fox from prairie dogs may be negated by the surrounding landscape fragmentation. The transformation of grassland into agricultural cropland has removed expanses of swift fox habitat (Kamler et al. 2003), and small colony sizes combined with annual geographical changes in colony placement may further influence swift fox presence. These impacts may describe the increasing trend of burrowing owls (USFS 2016a; Fig. 5a). Given that owls cue in on prairie dogs (Wunder and Bryan 2014), geographical changes in colony placement may not impact owl presence provided prairie dogs remain on the landscape. Additionally, owls will continue to breed on colonies impacted by plague providing burrows remain intact, vegetation does reach excessive heights, and prairie dogs re-colonize shortly after (Butts and Lewis 1982, Conrey 2010, Alverson and Dinsmore 2014). This suggests that
17


fragmented prairie dog colonies on the PNG have little impact on burrowing owls, further supporting the mesopredator release hypothesis (Crooks and Soule 1999, Prugh et al. 2009). Considering these fragmentation effects, mountain plover declines on prairie dog colonies may be responses to top-down control due to increased predation and/or selection of other suitable nesting habitat in the vicinity such as prescribed burns and fallow crop fields. Regardless, the conversion of shortgrass prairie to agricultural cropland potentially benefits mountain plovers and alludes to a case of landscape fragmentation that both negatively and positively impacts different levels of the potential trophic cascade.
Landscape Fragmentation on the TBNG
Landscape fragmentation impacts the northern mixed-grass prairie to a lesser extent. The soil structure is unsuitable for producing crops leaving cattle ranching as the primary agricultural practice (Fitzgerald et al. 1999). On the TBNG, black-tailed prairie dog colonies comprise large complexes upwards of-12,000 ha spanning public and privately-owned land. Annual mapping efforts dating to 1999 indicate that these expanses oscillate over time (Fig. 4), declining when epizootics of plague sweep through the landscape (USFS 2016b). Plague is less prevalent on an annual basis on TBNG and instead reflects patterns of density dependence (Cully et al. 2010). Prairie dog colonies were recovering from a plague epizootic at the start of the trend data in 2010 (USFS 2016b; Fig. 5b), so using plague as a driver of fragmentation on TBNG helps address fox, owl, and plover interactions.
Prairie dog colonies expanded across the grassland post 2010 (USFS 2016b; Fig. 5b), creating large, intact habitat patches for associated species. Systems with intraguild predation are frequently documented occurring on intact habitat patches (Robinson et al. 2014, Lonsinger et al.
18


2017), providing an explanation for the relationship between co-occurring foxes and owls. Foxes on the TBNG are found almost exclusively on prairie dog colonies and presence has increased following the prairie dog expansion (USFS 2016b). Burrowing owls are sparsely distributed across the prairie dog complexes and remain uncommon relative to the available habitat. If swift fox and burrowing owls are indicative of intraguild predation on TBNG, patterns in mountain plover trends follow suit: as fox occurrence increases, owls only minorly increase, and plovers increase (Fig. 5b). Plovers still risk predation by swift fox and other grassland predators; however, predation risk from owls is reduced. Thus, these potential cascading trophic interactions may be facilitated, in part, by the availability of intact patches of habitat created by prairie dogs.
Bottom-up Control
We have primarily focused on top-down control as a leading hypothesis to explain the trophic-ecology behind swift fox, burrowing owl, and mountain plover interactions on blacktailed prairie dog colonies. However, we cannot discount alternative ecological processes at work. Bottom-up control is characterized by primary producers or lower-level consumers driving changes to patterns in plant or wildlife populations and distributions at higher levels in the food-web (Elmhagen and Rushton 2007, Ripple et al. 2016; Fig. 3). Both the shortgrass and mixed-grass prairie communities are subject to changes in climate which change plant composition and structure (Polsky and Easterling III 2001). Here, an increase in percent precipitation may lead to taller grasses and forbs across prairie dog colonies changing the habitat conditions required by associated species. Vegetation structure on a colony is also influenced by plague, when prairie dogs are no longer present to clip the vegetation back to a shortgrass, bare-ground state (Augustine et al. 2008). Changes to plant composition and structure lead to changes in the insect
19


community (Whicker and Detling 1988), or may represent an absence of prairie dogs, which may then cascade up the food-web impacting species from the bottom up. We argue that it is likely never the case where only top-down or bottom-up control is at work, but instead both trophic-level processes are working concurrently to influence the biotic state of the community. Therefore, carefully considering all factors that may contribute to a bottom-up effect on a swift fox-burrowing owl-mountain plover interaction is critical for understanding all dynamics of the system.
Additional Influences from USFS Grassland Management Techniques
The PNG and TBNG manage for prairie dogs, and to varying degrees, associated species. Prairie dog management consists of both lethal and non-lethal management strategies. Lethal strategies on the PNG include non-regulated prairie dog shooting and poisoning of colonies that interfere with agricultural operations. TBNG has regulated a prairie dog shooting closure on colonies that are considered core areas for associated species. Local stakeholders successfully removed the shooting closure in recent history and poisoning of prairie dogs on the grassland has increased. Non-lethal prairie dog management includes structural (i.e. vegetation/fence) barriers on colony edges to prevent dispersal in any given direction on both grasslands, and translocation of prairie dogs from unsuitable areas to patches of core area habitat on the TBNG.
Plague mitigation in the form of application of an insecticide, deltamethrin, that targets fleas (order Siphonaptera - the vector of sylvatic plague; Eads and Biggins 2017) on prairie dogs has occurred on both the PNG and TBNG. Prescribed bums that create temporary habitat for mountain plover and swift fox have occurred on both USFS grasslands. Finally, grazing of domestic cattle occurs on pastures that overlap prairie dog colonies on both USFS grasslands. In areas where heavy grazing occurs outside of prairie dog colonies, bare ground habitat can result,
20


which may attract dispersing prairie dogs and breeding plovers. Grazing periods, stocking rates, and densities may have an impact on the likelihood of prairie dog and associated species occurrence and all yield varying levels of success. Thus, it is likely that management actions influence cascading tropic interactions by altering habitat composition and co-occurrence between species.
Conclusions
Trophic cascades occur in a diverse range of ecosystems and exist between flora and fauna at all trophic-levels (Paine 1980, Elmhagen and Rushton 2007, Ripple et al. 2016). Cascading trophic interactions in North American temperate grasslands remain poorly described despite a heavy management focus on many individual species in the system. The theoretical framework presented here provides a starting point for research on species interactions between swift fox, burrowing owl, and mountain plover on prairie-dog dominated rangeland communities. We add that there are many alternative, plausible hypotheses that we do not describe in detail in this manuscript. For example, mountain plover populations on both grasslands may be largely driven by nesting habitat availability, to include insect availability, and populations may reflect these factors more-so than direct or indirect influences from cooccurring species. And burrowing owl populations may increase with a direct correlation from nesting habitat availability independent from a decrease in swift fox populations. We also acknowledge that it remains unclear which life-stage predation on owls and/or plovers predominantly occurs.
We fully acknowledge that inferences made in this paper are anecdotal and maintain that future investigation into this system is needed. Trophic cascades are a driving force in community ecology that cannot be defined by any one mechanism. The presence or absence of
21


top-level consumers, hunting, disease, management strategies, agricultural production, habitat fragmentation, or deviations in resource availability, all which change temporally, contribute to the dynamics of trophic cascades. Other environmental variables not discussed here, such as climate, should also be considered. We propose targeted long-term multi-species occupancy research that combines the hypothesized mechanisms specified above with presence and cooccurrence of associated grassland species. This level of analysis is recommended for exploring these complex interactions moving forward, and this manuscript only serves as a prelude to this suggestion. Relationships between prairie dogs and associated species are highly interactive, so managing prairie dog colonies for viable populations of associated grassland wildlife is recommended from a conservation perspective. This promotes species richness, rangeland health, and diversity in shortgrass and mixed-grass prairies, and highlights opportunities for investigating multi-species interactions across all trophic-levels.
22


CHAPTER II
FACTORS INFLUENCING THE OCCURRENCE OF SWIFT FOX, BURROWING OWL, AND MOUNTAIN PLOVER ON BLACK-TAILED PRAIRIE DOG COLONIES
IN WYOMING
Introduction
Many species of conservation concern depend on patches of habitat within a broader landscape and fulfill a role in the ecological processes that drive their populations within these patches. The extent to which a given species can be associated with a habitat patch is influenced by variations in patch characteristics such as patch size, resource availability, human presence, disease, vegetation structure and diversity, and climate (Ye et al. 2013, Zuckerberg et al. 2018). Changes in these characteristics over space and time often drive habitat quality, which in turn may govern species-occupancy on a given habitat-patch. Inter-specific interactions may also lead to changes in occupancy rates on a patch for individual species (Grinde and Niemi 2016), suggesting that occupancy is not only a function of habitat quality, but also a function of the presence of co-occurring species. To quantify the effect of habitat quality and co-occurring species presence on occupancy, we can evaluate influencing characteristics at the patch level. Patch characteristics can be measured alongside repeated presence or absence surveys for one or more species to estimate occupancy probabilities that inform applied management decisions (Estevo et al. 2017). Management decisions directed towards meeting the habitat requirements for multiple co-occurring species work to mitigate declining animal populations (Sauer et al. 2013, White et al. 2013). Therefore, understanding the ecology behind population dynamics in a patchy environment is important for assessing metapopulations of wild animals, particularly cooccurring species of conservation concern.
23


Black-tailed prairie dogs (Cynomys ludovicianus) are a species of conservation concern at state and federal levels, and their colonies serve as individual patches of habitat in the Northern Great Plains for some species (Hoogland 2006, USDA 2017, WGFD 2017). Prairie dogs are small, social mammals that excavate complex tunnel systems that provide shelter for adults and offspring and a place to store food (Hoogland 2006). Above ground, prairie dog colonies are characterized by mounds of soil and clipped vegetation leaving a short-grass, bare ground landscape that allows for predator detection (Hoogland 2006). Prairie dogs are considered a keystone species that provide habitat for over 100 grassland associated species, and they serve a controversial role in rangeland and agricultural production by creating a bare-ground landscape that is undesirable for livestock producers who rely on grasses and forbs for cattle-forage (Hoogland 2006, Field et al. 2016). Prairie dog colonies change spatially and temporally across the landscape in response to habitat fragmentation, disease, lethal and non-lethal management, and climate, all of which contribute to changes in individual colony size, vegetation structure, and probability of occupancy by colony-associated species (Hoogland 2006, Johnson et al. 2011, Dinsmore and Smith 2010, Alverson and Dinsmore 2014). Collectively, the patch characteristics of a prairie dog colony support a complex system of ecological interactions.
Variations in landscape characteristics throughout the prairie dog’s range may change the dynamics of ecological interactions on a colony. For example, prairie dog colonies in the Northern Great Plains typically occur in mixed-grass prairie mosaics where a variety of habitat types intermix (Duchardt et al. 2018, Parker et al. 2019). Here, species that strongly benefit from prairie dogs are almost exclusively tied to colonies because they provide the only source of suitable habitat in the region (Parker et al. 2019). In this study, we address potential species interactions between three prairie dog-associates: swift fox (Vulpes velox), burrowing owl
24


{Athene cunicularia hypugaea), and mountain plover {Charadrius montanus) that strongly benefit from prairie dog colonies in the Northern Great Plains (Parker et al. 2019). Foxes, owls, and plovers are also listed as species of conservation concern at state and federal levels (USDA 2017, WGFD 2017).
Swift Fox: The swift fox is a small canid that frequently occupies shrub-free expanses across the Great Plains such as open fields, burned areas, and prairie dog colonies (Dark-Smiley and Keinath 2003, Kintigh and Anderson 2005). Prairie dog colonies provide denning opportunities and abundant prey resources for swift foxes to include prairie dogs, rabbits, mice, and ground nesting birds (Miller and Knopf 1993, Kintigh and Anderson 2005). Burrowing Owl: Burrowing owls are small raptors that nest and shelter in pre-excavated burrows (Poulin et al. 2011). In the Great Plains, owls have been documented on prairie dog colonies to respond to prairie dog alarm calls, suggesting that owls use prairie dogs for detecting predators (Bryan and Wunder 2014). Prairie dog colonies support abundant populations of ground nesting birds and insects; both prey items for burrowing owls (Conrey 2010). Mountain Plover: Mountain plovers are upland shorebirds that nest and forage on landscapes with relatively high bare-ground; a characteristic of prairie dog colonies (Augustine and Baker 2013). Prairie dog colonies are a strong source of habitat for breeding mountain plovers in the Northern Great Plains, where shortgrass prairies transition into mixed-grass/sagebrush steppe ecotones (Duchardt et al. 2018).
Parker et al. (2019) proposed that these species are involved in a complex web of predator and prey-interactions that reflect patterns resembling a trophic cascade. Because prairie dog colonies change spatially and temporally in the Northern Great Plains, and because foxes, owls, and plovers are strongly associated to these colonies (Dinsmore and Smith 2010, Alverson and Dinsmore 2014, Duchardt et al. 2018), we investigated whether the probability of occupancy
25


of a colony by a given species is influenced by the presence of co-occurring species at different trophic-levels, as well as colony size. Using individual prairie dog colonies as our sample unit, we used a hierarchical modeling approach (Royle and Kery 2007, Hobbs and Hooten 2015) to estimate single-species occupancy probabilities of prairie dog colonies by foxes, owls, and plovers under the following hypotheses (Box 1).
Box 1. Hypotheses regarding occupancy of black-tailed prairie dog colonies by swift fox, burrowing owl, and mountain plover:
1) Swift Fox:
a. Occupancy is informed by the combined presence absence of Burrowing Owls and Mountain Plovers.
b. Occupancy is informed by prairie dog colony size.
2) Burrowing Owl:
a. Occupancy is informed by the combined presence absence of Swift Fox and Mountain Plovers.
b. Occupancy is informed by prairie dog colony size.
3) Mountain Plover:
a. Occupancy is informed by the combined presence absence of Swift Fox and Burrowing Owls.
b. Occupancy is informed by prairie dog colony size.
Methods
Study area
We studied black-tailed prairie dogs, swift fox, burrowing owl, and mountain plover from 2016-2018 on the Thunder Basin National Grassland (TBNG) in the mixed-grass prairies of northeast Wyoming (Figure 1). The ~1600-km2 study area is bordered by Wyoming State Highways 450 to the north and 50 to the west, the Miller Hills to the south, and contiguous private land to the east. TBNG is comprised of a mixture of state owned or federally managed
26


US Forest Service (USFS) grassland and privately-owned ranches. Due to access limitations, this research was conducted on public lands only. TBNG is characterized as a northern mixed-grass prairie, consisting of ecotones of sage-brush (Artemisia sp.) steppe, ponderosa pine (Pinus ponderosa) and Rocky Mountain juniper (Juniperus scopulorum) forests, riverbeds lined with plains cottonwoods (Popidus deiloides), rocky outcroppings and escarpments, and patches of short-grass, bare ground prairie typically occupied by prairie dogs (USDA 2001, Duchardt et al. 2018). Grasses and forbs comprise the vegetation structure on prairie dog colonies, to include western wheat grass (Pascopyrum smithii), blue grama (Bouteloua gracilis), prairie junegrass (Koeleria macrantha), plains prickly pear (Opimtiapolyacantha), scarlet globemallow (Sphaeralcea coccinea), wooly plantain (Plantagopatagonica), and pepperweed. (Lepidinm sp.).
WY Hwy 450 * * » *
I Black-tailed Prairie Dog ' Colonies Sampled
i WY Hwy 50
~**«J!S*
• *
?r Hills A me .
A4! A £ * A 5! 5
[ i Kilometers
+*+ * r
Miller Hills /
# #
Thunder Basin National Grassland. Wyoming
Figure 1. Thunder Basin National Grassland (TBNG) is located in northeastern Wyoming. Sampled black-tailed prairie dog (Cynomys hidoviciamis) colonies, in black, ranged in size from 2 ha to 2767 ha with a mean of 456.64 ha and median of 146.5 ha (see Table 2b). Map created in Esri ArcMap 10.4.1.
27


Prairie dog colonies (patch sample unit)
Within the study area, we surveyed foxes, owls, and plovers exclusively on black-tailed prairie dog colonies (Figure 1). We used prairie dog colonies as our sample unit because each colony is a stand-alone habitat patch that requires a unique survey effort. Each colony was occupied by prairie dogs > 1 year during our three-year survey. All colonies occurred within TBNG boundaries and were all previously identified and mapped by USFS biologists or contracting biologists. Prairie dog colonies have been mapped in this landscape annually since 1999 to record spatial and temporal variation in colonies between years. Colonies were mapped by Trimble or Garmin GPS units to mark the physical outermost-boundary of active prairie dogs identified by the sight and sound of prairie dogs, evidence of active digging and fresh scat, and in-tact burrow systems. Burrows on the outermost-boundary that were collapsed or sealed off suggest prairie dog absence or inactivity and were excluded from mapping. On TBNG, prairie dog colonies are typically separated by stands of woody vegetation or deep drainages and gullies, and Ulev (2007) reports that burrows in the Northern Great Plains typically occur on <10% slope. Field observations and colony mapping aligned with this report and indicated that steep hillsides and rolling hills also divided colony boundaries on TBNG. Because prairie dog colonies on TBNG are annually susceptible to sylvatic plague (Cully Jr. et al. 2010, Eads et al. 2018), lethal and non-lethal management, and changes in climate, mapping efforts tracked temporal changes in colony characteristics. All mapping shapefiles used in this study were retrieved from USFS biologists or contracting biologists and were analyzed in ArcMap 10.4.1 to extract colony area for all colonies surveyed in each of our sample seasons.
Multi-species sampling
28


We recorded the presence or absence of swift fox, burrowing owl, and mountain plover from 2016-2018 within a breeding period survey window of May 1st and July 15th each year. Because breeding periods for all three species overlap and rearing of offspring carries into late July, we combined survey efforts for all three species into one multi-species survey and looked for all three species simultaneously. To account for occasions when species went undetected on a colony during a survey but remained present, we conducted repeat surveys on each colony three times within the survey window (MacKenzie et al. 2006). Surveys occurred from sunrise to ~11:00 a.m. and -5:00 p.m. to sundown to avoid the highest temperatures of the day. Surveys included driving weaving transects (i.e. zig-zag motion) across prairie dog colonies no more than 400m apart by truck or ATV, stopping every time a detection was made, or every -15 seconds to scan. Driving in weaving transects aided in flushing nesting plovers and provided a more exhaustive survey across the habitat patch for all species.
Our research rests on the assumption that the probability of occupancy remains the same at individual prairie dog colonies within a season and across all years for each species. A violation to this assumption occurs in the event of nest predation or when increased human presence results in mortality or the relocation of an adult-individual to a new habitat patch. We assumed that prairie dog colonies are independent and that detecting a species at one site is independent of detecting the same species at a different site (Bailey and Adams 2005,
MacKenzie et al. 2006). We monitored all detected species to identify whether an active nest or den existed at a site and then matched adult-individual(s) to a nest or den on the colony being sampled. Finally, we use colony size and associated species presence/absence as individual site-covariates to quantify differences between probabilities of occupancy across all sampled colonies.
29


Single-species hierarchical framework
We created single-species, robust-design occupancy models (MacKenzie et al. 2006) under a Bayesian hierarchical framework using JAGS (Package R2jags, R version 3.5.1, Royle and Kery 2007, Hobbs and Hooten 2015, Su and Yajima 2015) to investigate the impacts of associated-species presence and colony size on the probability of occupancy of prairie dog colonies by foxes, owls, and plovers. We created a set of models for each species that pools across all years and treats the three individual sampling seasons as a single-season survey. We pool across all years to increase sampling size instead of stratifying by individual year because our sparse data set prevents model convergence when analyzing between sampling seasons (see Appendix A for models stratified by year). We acknowledge that pseudoreplication (Hurlbert 1984) occurs under this approach; however, we continued with analysis to determine whether covariate-effects could be identified on occupancy probability estimates with an increased sample size. For all models, we generated prior distributions for Beta (distribution of covariates) and the probability of detection (P), using the logistic and uniform distributions, respectively (Northrup and Gerber 2018, Outhwaite et al. 2018). The following likelihood framework was used to model the logit probability of occupancy (F):
Likelihood:
logit(F[i]) = BetaO + Betal * XI[i] + Beta2 * X2[i].. .BetaN * XN[i]
Z[i] = bemouli(F[i])
Y_Sp[i] = binomial((Z[i] * P), J[i]) Equation 1.
Here, Xrepresents a patch-level covariate (colony size or co-occurring species presence/absence), i is the individual habitat patch (colony), Z represents the true state of occupancy, Y Sp refers to the total number of detections of a single species (Y SwiftFox,
30


YBurrow ingOw l, YMountainP lover), and ./is the total number of repeat visits in a season to habitat patch i (Equation 1). These models yield an estimate for the probability distributions of parameters psi (Â¥) and P, occupancy and detection, respectively. These parameters inform the proportion of prairie dog colonies occupied by each species, and the probability of detecting each species 1 or more times on a given colony. To further refine our estimate for W, we included a set of covariates as priori predictions and kept P constant. Because we are modeling sparse data, holding P constant lessens the number of parameters estimated and produces a more reliable result for W (Welsh et al. 2013), our parameter of interest.
Covariates to inform Psi (W)
Prairie dog colonies change spatially and temporally between years impacting the ecological characteristics of the sample unit and the species that are present. However, our data set only covered a three-year span (2016-2018) with small sample sizes in each year, so we pooled across years and assumed closure between all years. We included year as a covariate in a separate set of models and include the results and the list of non-converging models for this set in the appendices (Appendix A). To further refine our estimates of W, we evaluated covariate-effects from associated species presence and colony size.
Associated species presence: We surveyed for swift fox, burrowing owl, and mountain plover simultaneously on prairie dog colonies and each species received an individual presence score, 1, or an absence score, 0, based on the detection of that species during the survey. When estimating swift fox occupancy, we used the combined three-survey presence and absence scores for burrowing owl on each colony plus the presence and absence scores for mountain plover on each colony as a patch-site covariate (Table 1). When estimating burrowing owl occupancy, we used the combined three-survey swift fox presence/absence scores plus mountain plover
31


presence/absence scores as a patch-site covariate (Table 1). Finally, for estimating mountain plover occupancy, we used the combined three-survey presence/absence scores for swift fox plus presence/absence scores for burrowing owl (Table 1; see Robinson et al. 2014 for a similar approach). These covariates were used to address our initial hypotheses (Box 1) to explore proposed species-interactions that occur on prairie dog colonies (Parker et al. 2019). We used the presence/absence covariates for two species to measure an interaction-effect on occupancy for the third species because we are interested in species co-occurrence on any given colony (Table 1).
Prairie dog colony size: Patch size is frequently included in ecological models to estimate influences on species occurrence because it is often informative for determining the proportion of habitat necessary for a given species to meet its resource needs (Martinson et al. 2012, Shake et al. 2012). For example, prairie dog colonies in Phillips County, Montana, show a quadratic effect on colony size for probability estimates of occupancy of a colony by both burrowing owl and mountain plover (Dinsmore and Smith 2010, Alverson and Dinsmore 2014). This suggests that colony size plays an important role in the habitat selection process by an individual. We use colony area in hectares (ha) from the previous year (2015-2017) for each year surveyed (2016-2018, pooled across all years) as a covariate to estimate a patch-size effect and a quadratic patch-size effect on probability estimates of occupancy (Table 1). Using a previous-year patch-size approach is supported biologically because the success of a species on a given patch can predict the fidelity of that species to a habitat-patch in the following breeding season (Schmidt 2001). Raw data for colony size recorded in hectares was scaled ((area - mean) / standard deviation) for modeling in R2jags.
32


Table 1. Exhaustive list of models created to estimate the effect of co-occurring species presence and colony size on the probability that a prairie dog colony is occupied by swift fox, burrowing owl, and mountain plover. Here, letter a represents one of the three species, while letters b and c represent the combined three-survey presence/absence scores for the two-remaining species.
Models Definition
iPa(.) Pa (.) Species a probability of occupancy intercept model
UJa (area) Pa (.) Species a probability of occupancy with effect of colony size
U^a (area + area2) Pa (.) Species a probability of occupancy with quadratic effect of colony size
iPa {(b * c) + b + c) Pa (.) Species a probability of occupancy with interaction effect of species b and species c presence
^a {(b * c) + b + c + area) Pa (.) Species a probability of occupancy with interaction effect of species b and species c presence plus effect of colony size
^a ((b * c) + b + c + area + area2) Pa (.) Species a probability of occupancy with interaction effect of species b and species c presence plus quadratic effect of colony size
Results
We surveyed 24 colonies in 2016, 34 colonies in 2017, and 30 colonies in 2018 for a total sample size of 88 colonies over the three-year period (Table 2a). We recorded a total presence of swift fox on 25 colonies, burrowing owls on 34 colonies, and mountain plover on 42 colonies across all years (Table 2a). 36 of our sampled colonies were between 0 to 100 ha in size recorded from the previous year to sampling, representing 41% of the total colonies surveyed (Figure 2).
14 of our sampled colonies were >1000 ha in size recorded from the previous year to sampling, representing 16% of the total colonies surveyed (Figure 2). Colony size across all years ranged from 2 ha to 2767 ha with a mean of 456.64 ha and a median of 146.5 ha (Table 2b).
Table 2. a.) Number of colonies sampled each year, along with number of presence scores recorded for each species; b.) Range, mean, and median colony size (ha) pooled across years.
33


a. Year
2016 2017 2018 Total
# of prairie dog colonies sampled 24 34 30 88
# of recorded swift fox presence 8 15 2 25
# of recorded burrowing owl presence 7 17 10 34
# of recorded mountain plover presence 15 14 13 42
________Colony Size Pooled Across Years_____
Range (ha) Mean (ha) Median (ha)
2 - 2767 456.64 146.5
40 -i
Colony area (ha)
Figure 2. Distribution of n=88 black-tailed prairie dog colonies by size (hectares) on TBNG, representing the previous-year colony size (2015-2017) for each year sampled (2016-2018).
Our models pool data across all sampling seasons to increase our sample size to n = 88 (Table 2a) and test an effect of colony size and associated species presence on occupancy (Table 1). Swift Fox: Bayesian trace plots retrieved from R2jags output (Appendix A) and model estimates indicate that the best-converging model in this set was the intercept model that holds occupancy and detection constant (W = .649 ± .159, lower Cl =.379, upper Cl = .964', Table 3a).
34


We were unable to reliably estimate occupancy for the remaining models (Fis near or at 1.000; Table 3a). Burrowing Owl: Trace plots and model estimates indicate strong model convergence for the intercept model (F .475 . 070, lower Cl = .348, upper Cl = .623; Table 3b) and the
model testing a linear effect on colony size (F = .479 ± .078, lower Cl = .341, upper Cl = .644', Table 3b), and poor model convergence for all other models. Mountain Plover: Trace plots and model estimates also indicate strong model convergence for the intercept model (F = .604 ±
.073, lower Cl = .467, upper Cl = . 755; Table 3c) and the model testing a linear effect on colony size (F = .680 ± .101, lower Cl = .500, upper Cl = .897', Table 3c), and poor model convergence for all other models. In all three model sets, certain models produced probability estimates for occupancy that were smaller than the lower Cl (Table 3a-c); a further indication of sparse data.
Table 3. Analyzed model sets for a. Swift Fox, b. Burrowing Owl, and c. Mountain Plover. Covariates include colony size (area), a quadratic effect on colony size (area2), and the presence of swift fox (sf), burrowing owl (bo), and mountain plover (mp) and an interaction effect (*) between two of the species to estimate a probability of occupancy (F). Standard deviation and lower and upper 95% credible intervals are provided for the mean probability estimates of F.
a. Swift Fox Models SD L. Cl U. Cl
iy(.)P{.} .649 .159 .379 .964
•P (area) P (.) 1.000 .005 1.000 1.000
•P (area + area2) P (.) 1.000 .012 1.000 1.000
•P ((mp * bo) + mp + bo) P (.) .976 .111 .533 1.000
•P ((mp * bo) + mp + bo + area) P (.) .996 .047 1.000 1.000
IP ((mp * bo) + mp + bo + area + area2) P (.) .995 .047 1.000 1.000
b. Burrowing Owl Models SD L Cl U. Cl
IP(.)P{.) .475 .070 .348 .623
IP (area) P (.) .479 .078 .341 .644
•P (area + area2) P (.) .559 .325 .168 1.000
•P ((mp * sf) + mp + sf) P (.) .646 .270 .266 1.000
*P ((mp * sf) + mp + sf + area) P (.) .997 .041 1.000 1.000
*P ((mp * sf) + mp + sf + area + area2) P (.) .999 .017 1.000 1.000
c. Mountain Plover Models SD L Cl U. Cl
IP(.)P{.) .604 .073 .467 .755
IP (area) P (.) .680 .101 .500 .897
•P (area + area2) P (.) .597 .163 .296 .925
>P ((bo * sf) + bo + sf) P (.) .729 .222 .382 1.000
•P ((bo * sf) + bo + sf + area) P (.) .850 .180 .699 1.000
•P ((bo * sf) + bo + sf + area + area2) P (.) .282 .443 .000 1.000
35


We further evaluated the effect of colony size on the odds that a prairie dog colony is occupied by burrowing owls and mountain plovers by analyzing the coefficient estimates of our converging models. The mean logit Beta coefficient estimate for colony size on the probability of occupancy of a prairie dog colony by burrowing owl is .339 ± .306. We converted this estimate to odds (e2033 * 1 049) and report that for every 1-unit ha increase in colony size, the odds that a prairie dog colony is occupied by a burrowing owl increases by 1.403 ± 1.358. The mean logit Beta coefficient estimate for colony size on the probability of occupancy of a prairie dog colony by mountain plover is 2.033 ± 1.049. Thus, for every 1-unit ha increase in colony size, the odds that a mountain plover occupied a prairie dog colony increased by 7.637 ± 2.854.
Discussion
We are unable to report on the effect of associated-species presence or colony size on the probability that a prairie dog colony is occupied by a swift fox. We had n=25 swift fox presence records over a pooled sampling size of n=88 prairie dog colonies (Table 2a), which resulted in the failure of our models to properly converge. We know that swift foxes strongly benefit from prairie dog colonies in the northern Great Plains (Dark-Smiley and Keinath 2003, Parker et al. 2019), and studies suggest that prairie dog colonies provide a major source of food throughout the southern portion of the species range (Kintigh and Anderson 2005). Burrowing owls may represent a competing predator to swift fox, who together, share a prey resource in mountain plover (Parker et al. 2019). To this extent; however, we cannot say with any certainty that colony size or associated species presence influences the probability of occupancy of a swift fox on TBNG, and we allude that these inconclusive results reflect the sparse-nature of our data set.
A weak but positive effect of colony size influenced the probability that a prairie dog colony is occupied by a burrowing owl over a pooled sampling size, n=88. This finding is
36


consistent with studies occurring in Phillips County, Montana, which suggest that colony size is a determining factor in owl occupancy (Alverson and Dinsmore 2014). We had n=34 presence records for burrowing owl over our three-year sampling period, which is still a sparse representation within the data set (Table 2a). This is a likely explanation for the poor-model convergence in the remaining models within the set. We know that burrowing owls are strongly tied to prairie dog colonies where owl and prairie dog ranges overlap, and that burrow systems created by prairie dogs are an important component of a burrowing owl’s shelter, nesting, and brood rearing requirements (Poulin et al. 2011, Bryan and Wunder 2014). Additionally, the food resources that prairie dog colonies offer provide an excellent source of forage opportunity for owls (Conrey 2010). We cannot present with any certainty; however, that owls respond to the presence of swift fox or mountain plover on TBNG. Producing results that suggest any influence at all from an effect of associated species-presence on burrowing owl occupancy will require a more-robust data set.
A stronger positive effect of colony size was identified as an influencing factor on the probability that a prairie dog colony is occupied by a mountain plover over a pooled sampling size, n=88. Like burrowing owls, findings in Phillips County, Montana suggest that mountain plovers respond positively to changes in colony size (Dinsmore and Smith 2010). On TBNG, larger tracts of habitat-patches are likely more attractive to breeding plovers because these areas provide the only source of habitat across the landscape (Parker et al. 2019). We had n=42 presence records of mountain plover over our pooled data set, which was a sparse-enough number to prevent proper convergence of the remaining models in the set. We cannot say with certainty that associated-species presence influences the probability that a colony is occupied by a plover on TBNG, and we reiterate that a larger data set spanning multiple years and an
37


increased sample-unit size may be a better fit for occupancy-based analysis targeting multispecies systems of interest.
We cannot determine a threshold by which the size of a colony is maximized for predicting occupancy estimates. All models that estimate probability of occupancy as a function of a quadratic effect on colony size either poorly converged or produced an uninterpretable amount of noise associated with the Bayesian trace plots (Table 3a-c, S2 Appendix). We had an uneven distribution in size among our colonies sampled (Table 2b, Figure 2), which combined with a limited sample size, may explain our inability to produce a reliable set of estimates.
Research designs that investigate rates of occupancy typically contain extended periods of repeated annual surveying across many sampling units (MacKenzie et al. 2006). When data sets contain large sample sizes over several consecutive years, models can be constructed to estimate multiple parameters that define changes between years (MacKenzie et al. 2003). Because patch characteristics change between years, it makes sense biologically that the probability of associated-species occupying a colony would also change. We were unable to estimate a year effect on the probability of occupancy of prairie dog colonies by foxes, owls, and plovers due in part to sparse data. Extending this data set for several consecutive years may increase the likelihood of identifying a year effect on probability estimates and would be informative for targeting long-term management objectives. We note that the models that converged within our model-sets only contained estimates for a single covariate-effect (i.e. colony size); therefore, our justification for our results is that our sparse-data was unable to support an increased-number of parameters in our poorly-converging models.
Parker et al. (2019) proposed that foxes, owls, and plovers are involved in a multi-species interaction, where foxes control owls through intra-guild predation on TBNG. This would allow
38


for populations of a shared prey resource, mountain plover, to increase (Parker et al. 2019). US Forest Service trend data suggests that co-occurrence between foxes, owls, and plovers influenced individual species presence on both TBNG and Pawnee National Grassland in northeast Colorado (Parker et al. 2019). Our initial hypotheses address this interaction by suggesting that an effect on occupancy by one species occurs from the presence of other cooccurring species on TBNG. Our limited data-set prevented our ability to explore a species cooccurrence effect; however, these hypotheses should not be overlooked in future analyses. Our analysis does not confirm or deny that inter-specific interactions are a competing hypothesis when determining probability estimates for occupancy in patchy-habitats. Instead, our analysis suggests that sparse-data is a likely explanation for our negative results.
Robust-design multi-species occupancy models that address interactions between cooccurring species should target multi-year survey efforts with an increased number of patch-units sampled (Richmond et al. 2010, Robinson et al. 2014, Broms et al. 2016, Rota et al. 2016). Data sets of this nature allow for modeling the occupancy of one species conditional upon the occupancy of one or more species, rather than testing for an effect based on presence or absence as presented in this study. We also note that occupancy may not be defined by any-one covariate, and that influences from human presence, disease, vegetation structure and diversity, and climate necessarily must be considered in all ecological systems. We were able to identify effects of colony size on burrowing owl and mountain plover occupancy, and we can say that as colony size increases, the likelihood that a colony is occupied by owls or plover also increases.
However, colony size on TBNG is determined by lethal and non-lethal management, disease, and climate, and these factors influence the vegetation structure and composition. Taking these factors into consideration will be a necessary step if identifying interactions between species of
39


conservation concern in this complex system remains a target of interest. Such research will require the logistical resources necessary to invest in a long-term multi-species experiment in ever-changing patchy environment.


REFERENCES
Adler PB, HilleRisLambers J. 2008. The influence of climate and species composition on the population dynamics of ten prairie forbs. Ecology. 89:3049-3060.
Alverson KM, Dinsmore SJ. 2014. Factors affecting Burrowing Owl occupancy of prairie dog colonies. The Condor. 116:242-250.
Augustine DJ, Baker BW. 2013. Associations of grassland bird communities with black- tailed prairie dogs in the North American Great Plains. Conservation Biology. 27:324-334.
Augustine DJ, Dinsmore SJ, Wunder MB, Dreitz VJ, Knopf FL. 2008. Response of mountain plovers to plague-driven dynamics of black-tailed prairie dog colonies. Landscape Ecology. 23:689-697.
Augustine DJ, Matchett MR, Toombs TP, Cully Jr. JF, Johnson TL, Sidle JG. 2008.
Spatiotemporal dynamics of black-tailed prairie dog colonies affected by plague. Landscape Ecology. 23:255-267.
Augustine DJ, Skagen SK. 2014. Mountain plover nest survival in relation to prairie dog and fire dynamics in shortgrass steppe. Journal of Wildlife Management. 78:595-602.
Bailey L, Adams M. 2005. Occupancy models to study wildlife. USDI Geological Survey.
Patuxent Wildlife Research Center; Forest and Rangeland Ecosystem Science Center.
Broms KM, Hooten MB, Fitzpatrick RM. 2016. Model selection and assessment for multispecies occupancy models. Ecology. 97:1759-1770.
Bryan R, Wunder MB. 2014. Western burrowing owls {Athene cunicularia hypngaeci) eavesdrop on alarm calls of black-tailed prairie dogs (Cynomys ladovicianus). Ethology. 120: ISO-188.
Butts KO, Lewis JC. 1982. The importance of prairie dog towns to burrowing owls in Oklahoma. Oklahoma Cooperative Wildlife Research Unit. Oklahoma State University, Stillwater, Oklahoma.
Chi D. 2016. How a former Phoenix landfill became home for displaced burrowing owls. Audubon. (1 March 2018; www.audubon.org/news/how-former-phoenix-landfill-became-home-displaced-burrowing-owls).
Colorado Parks & Wildlife (CPW). 2016. Furbearer management report: 2015-2016 harvest year.
Colorado Parks & Wildlife (CPW). 2015. State wildlife action plan: A strategy for conserving wildlife in Colorado.
41


Conrey R. 2010. Breeding success, prey use, and mark-resight estimation of burrowing owls nesting on black-tailed prairie dog towns: Plague affects a non-susceptible raptor (Dissertation for Doctor of Philosophy). Digital Collections of Colorado, Colorado State University, Fort Collins, https://dspace.library.colostate.edu/handle/10217/39036.
Coupland RT. 1961. A reconsideration of grassland classification in the Northern Great Plains of North America. Journal of Ecology. 49:135-167.
Crooks KR, Soule ME. 1999. Mesopredator release and avifaunal extinctions in a fragmented system. Nature. 400:563-566.
Cully JF Jr, Johnson TL, Collinge SK, Ray C. 2010. Disease limits populations: Plague and black-tailed prairie dogs. Vector-Borne and Zoonotic Diseases. 10:7-15.
Dark-Smiley DN, Keinath DA. 2003. Species assessment for swift fox (Viilpes velox) in Wyoming. USD A Buraeu of Land Management. Wyoming State Office.
Desmond MJ, Savidge JA, Eskridge KM. 2000. Correlations between burrowing owl and blacktailed prairie dog declines: A 7-year analysis. Journal of Wildlife Management. 64:1067-1075.
Dinsmore SJ, Smith MD. 2010. Mountain plover responses to plague in Montana. Vector-Borne and Zoonotic Diseases. 10:37-45.
Dinsmore SJ, White GC, Knopf FL. 2005. Mountain plover population responses to black-tailed prairie dogs in Montana. Journal of Wildlife Management. 69:1546-1553.
Duchardt CJ, Porensky LM, Augustine DJ, Beck JL. 2018. Disturbance shapes avian communities on a grassland-sagebrush ecotone. Ecosphere. 9:e02483.
Eads DA, Biggins DE. 2017. Paltry past-precipitation: Predisposing prairie dogs to plague? Journal of Wildlife Management. 81:990-998.
Eads DA, Biggins DE, Bowser J, McAllister JC, Griebel RL, Childers E, Livieri TM, Painter C, Sterling-Krank L, Bly K. 2018. Resistance to deltamethrin in prairie dog (Cynomys htdoviciamts) fleas in the field and in the laboratory. Journal of Wildlife Diseases. 54:745-754.
Elmhagen B, Rushton SP. 2007. Trophic control of mesopredators in terrestrial ecosystems: Top-down or bottom-up? Ecology Letters. 10:197-206.
Estevo CA, Nagy-Reis MB, Nichols JD. 2017. When habitat matters: Habitat preferences can modulate co-occurrence patterns of similar sympatric species. PlosOne. 12:e0179489.
42


Field A, Sedivec K, Hendrickson J, Johnson P, Geaumont B, Xu L, Gates R, Limb R. 2016.
Effects of short-term cattle exclusion on plant community composition: Prairie dog and ecological site influences. Rangelands. 38:34-37.
Fitzgerald JA, Pashley DN, Pardo B. 1999. Bird conservation plan for the northern mixed-grass prairie (Physiographic Area 37). Partners in Flight (4 March 2018; https://www.partnersinflight.org/wp-content/uploads/2017/02/PA-37-Northern-Mixed-Grass-Prairie.pdf).
Grant TA, Shaffer TL, Madden EM, Nenneman MP. 2017. Contrasting nest survival patterns for ducks and songbirds in Northern mixed-grass prairie. Journal of Wildlife Management. 81:641-651.
Grinde AR, Niemi GJ. 2016. Influence of landscape, habitat, and species co-occurrence on occupancy dynamics of Canada Warblers. The Condor. 2016. 188:513-531.
Hobbs NT, Hooten MB. 2015. Bayesian models: A statistical primer for ecologists. Princeton University Press.
Hoogland JL. 2006. Conservation of the black-tailed prairie dog: Saving North America’s western grasslands. Island Press.
Hurlbert SH. 1984. Pseudoreplication and the design of ecological field experiments. Ecological Monographs. 54:187-211.
Jiang L, Joshi H, Patel SN. 2009. Predation alters relationships between biodiversity and temporal stability. The American Naturalist. 173:389-399.
Johnson TL, Cully JF Jr, Collinge SK, Ray C, Frey CM, Sandercock BK. 2011. Spread of plague among black-tailed prairie dogs is associated with colony spatial characteristics. The Journal of Wildlife Management. 75:357-368.
Kamler JF, Ballard WB, Fish EB, Lemons PR, Mote K, Perchellet CC. 2003. Habitat use, home ranges, and survival of swift foxes in a fragmented landscape: Conservation implications. Journal of Mammalogy. 84:989-995.
Karki SM, Gese EM, Klavetter ML. 2007. Effects of coyote population reduction on swift fox
demographics in Southeastern Colorado. Journal of Wildlife Management. 71:2707-2718.
Kintigh KM, Anderson MC. 2005. A den-centered analysis of swift fox (Viilpes velox) habitat characteristics in Northeastern New Mexico. The American Naturalist. 154:229-239.
Kitchen AM, Gese EM, Schauster ER. 1999. Resource partitioning between coyotes and swift foxes: Space, time, and diet. Canadian Journal of Zoology. 77:1645-1656.
43


Lauenroth WK, Burke IC. 2008. Ecology of the shortgrass steppe: A long term perspective. Oxford University Press, Inc.
Lauenroth WK, Burke IC, Gutmann, M.P. 1999. The structure and function of ecosystems in the central North American grassland region. Great Plains Research. 9:223-259.
Lindeman RL. 1942. The trophic-dynamic aspect of ecology. Ecology. 23:399-417.
Lonsinger RC, Gese EM, Bailey LL, Waits LP. 2017. The roles of habitat and intraguild predation by coyotes on the spatial dynamics of kit foxes. Ecosphere.
8(3):e01749.10.1002/ecs2.1749.
Macias-Duarte A, Conway CJ. 2015. Distributional changes in the western burrowing owl
(.Athene cunicularia hypugaea) in North America from 1967 to 2008. Journal of Raptor Research. 49:75-83.
MacKenzie DI, Nichols ID, Hines JE, Knutson MG, Franklin AB. 2003. Estimating site
occupancy, colonization, and local extinction when a species is detected imperfectly. Ecology. 84:2200-2207.
MacKenzie DI, Nichols JD, Royle JA, Pollock KH, Bailey LL, Hines JE. 2006. Occupancy estimation and modeling: Inferring patterns and dynamics of species occurrence. Academic Press.
Manning JA. 2011. Factors affecting detection probability of burrowing owls in southwest agroecosystem environments. Journal of Wildlife Management. 75:1558-1567.
Martin DJ, White GC, Pusateri FM. 2007. Occupancy rates by swift fox (Vulpes velox) in eastern Colorado. The Southwestern Naturalist. 52:541-551.
Martinson HM, Fagan WF, Denno RF. 2012. Critical patch sizes for food-web modules. Ecology. 93:1779-1786.
Miller BJ, Knopf FL. 1993. Growth and survival of mountain plovers. Journal of Field Ornithology. 64:500-506.
Nicholson KL, Ballard WB, McGee BK, Surles J, Kamler JF, Lemons PR. 2006. Swift fox use of black-tailed prairie dog towns in Northwest Texas. The Journal of Wildlife Management. 70:1659-1666.
Northrup JM, Gerber BD. 2018. A comment on priors for Bayesian occupancy models. PlosOne. 13:e0192819.
Outhwaite CL, Chandler RE, Powney GD, Collen B, Gregory RD, Issac NJB. 2018. Prior
specification in Bayesian occupancy modeling improves analysis of species occurrence data. Ecological Interactions. 93:333-343.
44


Paine RT. 1980. Food webs: Linkage, interaction strength and community infrastructure.
Journal of Animal Ecology. 49:666-685.
Parker RP, Duchardt CJ, Dwyer AM, Painter C, Pierce AK, Michels TJ, Wunder MB. 2019. Trophic ecology warrants multi-species management in a grassland setting: Proposed species interactions on black-tailed prairie dog colonies. Rangelands. ***. ***.
Plumb RE, Knopf FL, Anderson SH. 2005. Minimum population size of mountain plovers breeding in Wyoming. The Wilson Bulletin. 117:15-22.
Polis GA, Holt RD. 1992. Intraguild predation: The dynamics of complex trophic interactions. Trends in Ecology & Evolution. 7:151-154.
Polsky C, Easterling III WE. 2001. Adaptation to climate variability and change in the US Great Plains: A multi-scale analysis of Ricardian climate sensitivities. Agriculture, Ecosystems & Environment. 85:133-144.
Poulin RG, Todd LD, Haug EA, Millsap BA, Martell MS. 2011. Burrowing owl {Athene
cnnicnlaria), version 2.0. In The Birds of North America (Poole AF, Editor). Cornell Lab of Ornithology, Ithaca, NY, USA. https://doi.org/10.2173/bna.61.
Prugh LR, Stoner CJ, Epps CW, Bean WT, Ripple WJ, Laliberte AS, Brashares JS. 2009. The rise of the mesopredator. BioScience. 59:779-791.
Ramsdell PC, Sorice MG, Dwyer AM. 2016. Using financial incentives to motivate conservation of an at-risk species on private lands. Environmental Conservation. 43:34-44.
Richmond OMW, Hines JE, Beissinger SR. 2010. Two-species occupancy models: A new parameterization applied to co-occurrence of secretive rails. Ecological Applications. 20:2036-2046.
Ripple WJ, Estes JA, Schmitz OJ, Constant V, Kaylor MJ, Lenz A, Motley JL, Self KE, Taylor DS, Wolf C. 2016. What is a trophic cascade? Trends in Ecology & Evolution. 31:842-849.
Robinson QH, Bustos D, Roemer GW. 2014. The application of occupancy modeling to evaluate intraguild predation in a model carnivore system. Ecology. 95:3112-3123.
Rota CT, Ferreira MAR, Kays RW, Forrester TD, Kalies EL, McShea WJ, Parsons AW, Millspaugh JJ. 2016. A multispecies occupancy model for two or more interacting species. Methods in Ecology and Evolution. 7:1164-1173.
Royle JA, Kery M. 2007. A Bayesian state-space formulation of dynamic occupancy models. Ecology. 88:1813-1823.
45


Samson F, Knopf F, 1994. Prairie Conservation in North America. BioScience. 44:418-421.
Sauer JR, Blank PJ, Zipkin EF, Fallon JE, Fallon FW. 2013. Using multi-species occupancy models in structured decision making on managed lands. The Journal of Wildlife Management. 77:117-127.
Sauer JR, Link WA, Fallon JE, Pardieck KL, Ziolkowski Jr. DJ. 2013. The North American breeding bird survey 1966-2011: Summary analysis and species accounts. North American Fauna. 79:1-32.
Schmidt KA. 2001. Site fidelity in habitats with contrasting levels of nest predation and brood parasitism. Evolutionary Ecology Research. 3:633-648.
Shake CS, Moorman CE, Riddle JD, Burchell II MR. 2012. Influence of patch size and shape on occupancy by shrubland birds. The Condor. 114:268-278.
Smith MD, Conway CJ, Ellis LA. 2005. Burrowing owl nesting productivity: A comparison
between artificial and natural burrows on and off golf courses. Wildlife Society Bulletin. 33:454-462.
Soule ME, Estes JA, Miller B, Honnold DL. 2005. Strongly interacting species: Conservation policy, management, and ethics. BioScience. 55:168-176.
Stratman M. 2017. Status of swift fox in eastern Colorado. Colorado Parks & Wildlife.
Stukel ED. 2017. Swift fox conservation team: Report for 2015-2016. Wildlife Division Report No. 2017-04, SD Department of Game, Fish and Parks, Pierre, SD, USA.
Su YS, Yajima M. 2015. R2jags: Using R to run ‘JAGS’. R package version 0.5-7. https://CRAN.R-proiect.org/packageMl2iags.
Thiele JP, Bakker KK, Dieter CD. 2013. Multiscale nest site selection by burrowing owls in western South Dakota. The Wilson Journal of Ornithology. 125:763-774.
Thompson CM, Augustine DJ, Mayers DM. 2008. Swift fox response to prescribed fire in shortgrass steppe. Western North American Naturalist. 68:251-256.
Ulev E. 2007. Cynomys ludovicianus. In: Fire Effects Information System, [Online], U.S. Department of Agriculture, Forest Service, Rocky Mountain Research Station, Fire Sciences Laboratory (Producer), www.fs.fed.us/database/feis/mammal/cvlu/all.html.
USDA Forest Service. 2017. FSM 2600 - Wildlife, fish, and sensitive plant habitat management. Chapter 2670 - Threatened, endangered and sensitive plants and animals. Forest Service Manual. Rocky Mountain Region, Denver, CO.
46


USDA Forest Service. 2001. Land and resource management plan for the Thunder Basin National Grassland: Medicine Bow-Routt National Forest Rocky Mountain Region. Medicine Bow-Routt National Forests and Thunder basin National Grassland.
U.S. Fish and Wildlife Service (USFWS). 2002. Species assessment and listing priority assignment form. US Fish and Wildlife Service website:
https://www.fws.gov/southdakotafieldoffice/BTPD RO Cand.%20FINAL 12Aug04.pdf U.S. Forest Service (USFS). 2016a. Pawnee National Grassland.
U.S. Forest Service (USFS). 2016b. Thunder Basin National Grassland. Douglas Ranger District.
Virchow DR, Hygnstrom SE. 2002. Distribution and abundance of black-tailed prairie dogs in the great plains: A historical perspective. Great Plains Research. 12:197-218.
Webb CT, Brooks PB, Gage KL, Antolin MF. 2006. Classic flea-borne transmission does not
drive plague epizootics in prairie dogs. Proceeding of the National Academy of Sciences of the United States of America. 103:6236-6241.
Welsh AH, Lindenmayer DB, Donnelly CF. 2013. Fitting and interpreting occupancy models. PlosOne. 8:e52015.
Whicker AD, Detling JK. 1988. Ecological consequences of prairie dog disturbances.
BioScience 38:778-785.
White AM, Zipkin EF, Manley PN, Schlesinger MD. 2013. Conservation of avian diversity in the Sierra Nevada beyond a single-species management focus. PlosOne. 8:e63088.
Williford D, Woodin MC, Skoruppa MK. 2009. Factors influencing selection of road culverts as winter roost sites by western burrowing owls. Western North American Naturalist. 69:149-154.
Wunder MB, Knopf FL, Pague CA. 2003. The high-elevation population of mountain plovers in Colorado. The Condor. 105:654-662.
Wyoming Game & Fish Department (WGFD). 2017. State Wildlife Action Plan.
Ye X, Skidmore AK, Wang T. 2013. Within-patch habitat quality determines the resilience of specialist species in fragmented landscapes. Landscape Ecology. 28:135-147.
Zuckerberg B, Ribic CA, McCauley LA. 2018. Effects of temperature and precipitation on
grassland bird nesting success as mediated by patch size. Conservation Biology. 32:872-882.
47


APPENDIX
A. Chapter II Additional Tables and Figures
Table A I. Exhaustive list of models created to estimate a year effect, along with the effect of co-occurring species presence and colony size on the probability that a prairie dog colony is occupied by swift fox, burrowing owl, and mountain plover. Here, letter a represents one of the three species, while letters b and c represent the combined three-survey presence/absence scores for the two-remaining species.
Models Definition
fa (.) Pa {.) Species a probability of occupancy intercept model
fa (year) Pa (.) Species a probability of occupancy with effect of year
fa (area + year) Pa (.) Species a probability of occupancy with effect of colony size plus year
fa (area + area2 + year) Pa (.} Species a probability of occupancy with quadratic effect of colony size plus year
f a ((b * c) + b + c + year) Pa (.) Species a probability of occupancy with interaction effect of species b and species c presence plus year
fa ((b * c) + b + c + area + year) Pa (.) Species a probability of occupancy with interaction effect of species
b and species c presence plus effect of colony size plus year
fa ((b * c) + b + c + area + area2 + year) Pa (.) Species a probability of occupancy with interaction effect of species b and species c presence plus quadratic effect of colony size plus year
48


Table A II. Analyzed model sets for a. Swift Fox, b. Burrowing Owl, and c. Mountain Plover. All models contain a year covariate. Other covariates included are colony size (area), a quadratic effect on colony size (area2), and the presence of swift fox (sf), burrowing owl (bo), and mountain plover (mp) and an interaction effect (*) between two of the species to estimate a probability of occupancy {Â¥). Standard deviation and Lower and upper 95% credible intervals are provided for the mean probability estimates of W.
a. Swift Fox Models IP SD L. Cl U. Cl
Ui(.)P(.) .649 .159 .379 .964
If (year) P (.) .499 .498 .000 1.000
if (area + year) P (.) .499 .498 .000 1.000
If (area + area2 + year) P (.) .498 .497 .000 1.000
if ((mp * bo) + mp + bo + year) P (.) .502 .497 .000 1.000
if ((mp * bo) + mp + bo + area + year) P (.) .500 .498 .000 1.000
if ((mp * bo) + mp + bo + area + area2 + year) P (.) .501 .497 .000 1.000
b. Burrowing Owl Models IP SD L. Cl U. Cl
.475 .070 .348 .623
if (year) P (.) .501 .497 .000 1.000
if (area + year) P (.) .497 .498 .000 1.000
if (area + area2 + year) P (.) .500 .498 .000 1.000
if ((mp * sf) + mp + sf + year) P (.) .500 .497 .000 1.000
*4J ((mp * sf) + mp + sf + area + year) P (.) .502 .498 .000 1.000
if ((mp * sf) + mp + sf + area + area2 + year) P (.) .507 .498 .000 1.000
c. Mountain Plover Models IP SD L. Cl U. Cl
W(.)P(.) .604 .073 .467 .755
if (year) P (.) .495 .497 .000 1.000
If (area + year) P (.) .503 .498 .000 1.000
if (area + area2 + year) P (.) .498 .498 .000 1.000
If ((bo * sf) + bo + sf + year) P (.) .499 .497 .000 1.000
if ((bo * sf) + bo + sf + area + year) P (.) .497 .497 .000 1.000
If ((bo * sf) + bo + sf + area + area2 + year) P (.) .503 .498 .000 1.000
49


Figure A I a-f. Bayesian trace plots: Swift Fox
a. Model: Y (.)P(.)
Trace of betaO
Trace of deviance
Iterations
Iterations
Trace of p Trace of psi
Iterations
Iterations


100 110 120 130 140 150 0 200 400 600 800 1000
b. Model: Y (area) P (.)
Trace of betaO Trace of betal
Iterations
Iterations
Trace of deviance
Iterations
Trace of psiO
10000 20000 30000 40000 50000
Trace of p
Iterations
Iterations


0.10 0.15 0.20 0.25 -200
c. Model: Y (area + area2) P (.)
Trace of betaO
Iterations
Trace of betal
Iterations
Trace of beta2
Iterations
Trace of deviance
Trace of p
Iterations
Trace of psiO
10000 20000 30000 40000 50000
Iterations
52


0.10 0.15 0.20 0.25 -500 0 500 1000
d. Model: Y ((mp * bo) + mp + bo) P (.)
Trace of betaO
Iterations
Trace of betal
Iterations
Trace of beta2
Iterations
Trace of deviance
Iterations
Trace of p
Iterations
Trace of psiO
Iterations
53


125 135 145 -500 0 500 1000
e. Model: Y ((mp * bo) + mp + bo + area) P (.)
Trace of betaO
Trace of betal
Iterations
Trace of beta2
30000
Iterations
Trace of deviance
Trace of beta3
Iterations Trace of p
30000
Iterations
Trace of psiO
10000 20000 30000 40000 50000
Iterations
54


0.10 0.15 0.20 0.25 -200 0 200 600 1000 -500 0 500 1000 0 500 1000 1500
f. Model: Y ((mp * bo) + mp + bo + area + area2) P (.)
Trace of betaO Trace of betal
Trace of beta2
Trace of beta3
Iterations
Trace of beta4
Trace of deviance
Iterations
Iterations
Trace of p
Iterations
Trace of psiO
10000 20000 30000 40000 50000
Iterations
55


Figure A II a-f. Bayesian trace plots: Burrowing Owl
a. Model: Y (.)P(.)
Trace of betaO Trace of deviance
Trace of p Trace of psi
Iterations
Iterations


b. Model: Y (area) P (.)
Trace of betaO
Iterations
Trace of deviance
Iterations Trace of psiO
Iterations
Trace of betal
Iterations Trace of p
Iterations
57


c. Model: Y (area + area2) P (.)
Trace of betaO
10000 20000 30000 40000 50000
Iterations
Trace of beta2
10000 20000 30000 40000 50000
Iterations Trace of p
Iterations
Trace of betal
10000 20000 30000 40000 50000
Iterations
Trace of deviance
Iterations
Trace of psiO
p
10000 20000 30000 40000 50000
Iterations
58


500 1000
d. Model: Y ((mp * sf) + mp + sf) P (.)
Trace of betaO
Iterations
Trace of beta2
T
10000 20000 30000 40000 50000
Iterations
Trace of p
10000 20000 30000 40000 50000
Iterations
Iterations
Trace of deviance
10000 20000 30000 40000 50000
Iterations
Trace of psiO
10000 20000 30000 40000 50000
Iterations
59


160 170 180 190 200 210 -500 0 500 1000 0 200 400 600 800 1000
e. Model: Y ((mp * sf) + mp + sf + area) P (.)
Trace of betaO
Iterations
Trace of beta2
Iterations
Trace of deviance
Iterations
Trace of psiO
Trace of betal
Iterations
Trace of beta3
10000 20000 30000 40000 50000
Iterations Trace of p
Iterations
Iterations


f. Model: Y ((mp * sf) + mp + sf + area + area2) P (.)
Trace of betaO
Iterations
Trace of beta2
Iterations
Trace of beta4
o ---------------------------------
o
o -
o i
10000 20000 30000 40000 50000
Iterations
Trace of p
Iterations
Trace of betal
Iterations
Trace of beta3
o
o - _
10000 20000 30000 40000 50000
Iterations
Trace of deviance

10000 20000 30000 40000 50000
Iterations
Trace of psiO
Iterations
61


Figure A III a-f. Bayesian trace plots: Mountain Plover
a. Model: Y (.)P(.)
Trace of betaO
Trace of p
Trace of deviance
10000 20000 30000 40000 50000
Iterations
Trace of psi
Iterations
Iterations


200
b. Model: Y (area) P (.)
10000 20000 30000 40000 50000
Iterations
Trace of deviance
10000 20000 30000 40000 50000
Iterations
Trace of psiO
Iterations
Trace of betal
Trace of p
63


c. Model: Y (area + area2) P (.)
Trace of betaO
Iterations
Trace of beta2
Iterations
Trace of p
Iterations
Trace of betal
Iterations
Trace of deviance
10000 20000 30000 40000 50000
Iterations
Trace of psiO
Iterations
64


500 1000
d. Model: Y ((bo * sf) + bo + sf) P (.)
Trace of betaO
Iterations
Trace of beta2
Iterations
Trace of p
Iterations
Trace of betal
Iterations
Trace of deviance
Iterations Trace of psiO
Iterations
65


200 400 600 800
e. Model: Y ((bo * sf) + bo + sf + area) P (.)
Trace of betaO
Iterations
Trace of beta2
10000 20000 30000 40000 50000
Iterations
Trace of deviance
T
10000 20000 30000 40000 50000
Iterations
Trace of psiO
10000 20000 30000 40000 50000
Iterations
Trace of betal
Iterations
Trace of beta3
10000 20000 30000 40000 50000
Iterations
Trace of p
10000 20000 30000 40000 50000
Iterations
66


200 600 1000 -200 0 200 400 600 800 0 200 400 600
f. Model: Y ((bo * sf) + bo + sf + area + area2) P (.)
Trace of betaO
Iterations
Trace of beta2
10000 20000 30000 40000 50000
Iterations
Trace of beta4
10000 20000 30000 40000 50000
Iterations Trace of p
Iterations
Trace of betal
Iterations
Trace of beta3
10000 20000 30000 40000 50000
Iterations
Trace of deviance
10000 20000 30000 40000 50000
Iterations
Trace of psiO
Iterations
67


Full Text

PAGE 1

PATTERNS OF OCCURRENCE AND CO OCCURRENCE FOR SWIFT FOX ( VULPES VELOX ), WESTERN BURROWING OWL ( ATHENE CUNICULARIA HYPUGAEA ), AND MOUNTAIN PLOVER ( CHARADRIUS MONTANUS ) ON BLACK TAILED PRAIRIE DOG (CYNOMYS LUDOVICIANUS ) COLONIES: A TREND DATA SUMMARY AND A HI ERARCHICAL OCCUPANCY ANALYSIS By RYAN ANDREW PARKER B.S., University of Wyoming, 2015 A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Master of Sci en ce Biology Program 201 9

PAGE 2

ii This thesis for the Master of Science degree by Ryan Andrew Parker has been approved for the Biology Program by Michael B. Wunder, Chair Diana F. Tomback Laurel M. Hartley Angela M. Dwyer Date: Ma y 18 , 2019

PAGE 3

iii Parker, Ryan Andrew (M.S., Biology Program) Patterns of Occurrence and Co Occurrence for Swift Fox ( Vulpes velox ), Western Burrowing Owl ( Athene cunicularia hypugaea ), and Mountain Plover ( Charadrius montanus ) on Black tailed Prairie Dog ( Cynomys ludovicia nus ) Colonies : A Trend Data Summary and a Hierarchical Occupancy Analysis Thesis directed by Associate Professor Michael B. Wunder A BSTRACT Swift fox ( Vulpes velox ), Burrowing Owl ( Athene cunicularia ) and Mountain Plover ( Charadrius montanus ) are three s pecies of con servation concern who all co occur on black tailed prairie dog ( Cynomys ludovicianus ) colonies across the Great Plains of North America. In this thesis, we examine the relationships between these species by 1) proposing an investigation into f ood web inter actions between foxes, owls, and plovers found specifically on prairie dog colonies, and 2) examining influences on occurrence of foxes, owls, and plovers on prairie dog colonies. Literature to date heavily focuses on each of these species individually, bu t a knowledge gap exists when considering multi species ecology in this system. Exploring h ow multiple co occurring species interact in Great Plains systems will provide a better understanding of the ecology across this fragile landscape. Chapter One explo res the possibility of foxes, owls, and plovers interacting across trophic levels. Trophic cascades occur when flora and fauna directly and/or indirectly influence co occurring species populations at different levels of the food chain, and North American t emperate grasslands provide an interesting case study to research these relationships. We b riefly

PAGE 4

iv define trophic cascades in terrestrial systems and explore the potential for a cascading trophic interaction between grassland associated swift fox, burrowing owl, and mountain plover, three rangeland species of conservation concern, on black tailed prairie dog colonies using two U.S. Forest Service data sets. Historic patterns of occurrence and co occurrence suggest top down control governs the spatiotemporal distribution patterns of the three species and may be influenced by habitat fragmentation a nd management actions. Managing for interactive, multi trophic communities requires the identification of species interactions and the mechanisms that drive them. Lo ng term multi species occupancy research combined with hypothesized driving mechanisms and the co occurrence of associated grassland species is recommended for addressing these complex interactions moving forward. Chapter Two focuses on the occurrence of f oxes, owls, and plovers on prairie dog colonies. Co occurring species often benefit from the same patches of habitat, and in some cases may be involved in complex interactions driven by patch characteristics. These characteris tics can include but are not l imited to patch size and resource availability, which can be coupled with the presence or absence of multiple co occurring species to evaluate probability estimates of occupancy. P rairie dog colonies serve as patches of habita t and swift fox, burrowing owl , and mountain plover co occur on these colonies benefiting from the shelter and food resources provided. In this chapter , we derive probability estimates for occupancy of prairie dog colonies by f ox es , owl s , and plover s in ea stern Wyoming. We hypothesize that co occurring species presence or absence of two species on a colony influences the occupancy of the third species, and that colony size also drives occurrence. We use hierarchical modeling to develop single species occupa ncy models for each species ac ross n=88 prairie dog colonies pooled across three sampling seasons (2016 2018). We were unable to identify effects on probability estimates of

PAGE 5

v occupancy for foxes, likely because sparse data prevented model convergence. For o wls and plovers, we identified an effect of colony size on occupancy. For every 1 unit ha increase in colony size, the odds that a colony was occupied by an owl and plover increased by 1.403 ± 1.358 and 7.637 ± 2.854, respectively. Our sparse data set like ly prevented proper model conv ergence for the remaining owl and plover models. Our results highlight that occupancy driven analyses require more robust data sets spanning multiple years of sampling and an increased number of sample units. We cannot confirm nor deny that foxes, owls, an d plovers are involved in complex interactions within a patchy environment, and our results warrant a long term, multi species experiment in this system. The form and content of this abstract are approved. I recommend its publ ication. Approved: Michael B. Wunder

PAGE 6

vi A CKNOWLEDGEMENTS A huge thank you goes out to my advisor, Dr. Michael B. Wunder, for his guidance on scientific literacy and writing , experimental design, ecological data analysis, career building, critical and independent thinking, and tim e management. A huge thank you to Dr. Diana F. Tomback and Dr. Laurel M. Hartley for their guidance and hours of discussion on my system of study, scientific methods, navigating a M.S. thesis, and scientific writi ng and communication . Another huge thank yo u to Angela M. Dwyer with Bird Conservancy of the Rockies for guidance on my system of study, scientific writing and communication, funding for field work, and for seeing the importance of my graduate work as it a pplies to the conservation and ecology of t he North American Great Plains. I would also like to give a huge thank you to Cristi Painter with the U.S. Forest Service for being instrumental in developing, supporting, and seeing this graduate project through till the end on USFS lands in eastern Wyomi ng, and also for being a mentor and amazing friend. To my fellow graduate student friends, thank you for countless hours of discussion, critical thinking, constructive advice, and mental support throughout the cou rse of my project. In particular, I would l ike to thank Ben Lagasse, Sara St. Onge, Tyler Michels, Alli Pierce, Scott Yanco, Amber Carver, Libby Pansing, Michelle Deprenger Levin, Andrew McDevitt, Katie Kilpatrick, David Schutt, Marianne Davenport, Mac Cal vert, and Katherine Fu. To Timothy Fosmo an d Joseph Alfonso, a huge thank you for help ing collect data in the field for my project. An enormous thank you goes out to my friends and family who provided me with much needed mental and emotional support for th ly, my biggest thank you goes out to my partner and number one fan, Christopher Gilbert . All wildlife research methods were approved by the University of Colorado Institutional Animal Care and Use Committee (IACUC Protocol # 92014(05)1C )

PAGE 7

vii TABLE OF CONTENTS CHAPTER I. TROPHIC ECOLOGY WARRANTS MULTI SPECIES MANAGEMENT IN A GRASSLAND SETTING: PROPOSED SPECIES INTERACTIONS ON BLACK TAILED PRAIRIE DOG COLONIES 1 1 . . . 2 . .. 4 .. 6 ... .. 9 Trophic Interactions on the Pawnee and Thunder Basin National 9 Hypotheses Concerning Trophi c Ecology on Pawnee and Thunder Basin National Grasslands ... 1 2 Additional Influences from USFS Grassland Management Techniques ...2 0 2 1 II. FACTORS INFLUENCING THE OCCURRENCE OF SWIFT FOX, BURROWING O WL, AND MOUNTAIN PLOVER ON BLACK TAILED PRAIRIE DOG COLONIES IN WYOMING ... 2 2 22 2 6 33 .. 36

PAGE 8

viii .. 41 APPENDIX 4 8

PAGE 9

1 C HAPTER I T ROPHIC E COLOGY WARRANTS M ULTI S PECIES M ANAGEMENT IN A GRASSLAND SETTING : PROPOSED SPECIES INTERACTIONS ON B LACK TAILED P RAIRIE D OG C OLONIES Introduction North American temperate grasslands support high levels of species richness and diversity and are an important resource for agricultural production , so research and appropriate management for these rangeland systems necessarily must focus on the interactio ns be tween species, rather than on single species populations (Samson and Knopf 1994, Augustine and Baker 2013, Grant et al. 2017) . Because species interaction paradigms are crucial for management in these systems, w e present a theoretical framework to add ress a broad issue: that cascading trophic interactions play a critical role in structuring populations of concern in grasslands. As a case study, we examine the hypothesis that co occurring swift fox ( Vulpes velox ), western burrowing owl ( Athene cunicular ia hy pugaea ) and mountain plover ( Charadrius montanus ) are involved in a cascading trophic interaction on b lack tailed prairie dog ( Cynomys ludovicianus ) colonies o n North American temperate grasslands. Trophic cascades among these species remain undocumen ted and are worthy of investigation for effective multi species management in these communities of conservation concern . We begin by introducing North American temperate grasslands and follow with short species accounts for p rairie dogs, foxes, owls, and p lovers . We then briefly define trophic interactions, proceeded by a case study on the Pawnee National Grassland (PNG) in Colorado and Thunder Basin National Grassland (TBNG) in Wyoming . Finally, we hypothesize potentially inf luencing mechanisms in a fox ow l plover

PAGE 10

2 interaction . We fully acknowledge that i nferences based on the trend data and literature cited in this paper are anecdotal and statistically insufficient and maintain that future investigation into this system is nee ded. Collectively, this paper s erves as a pro posal for future trophic level research in this interactive communit y . North American T emperate G rasslands North American t emperate grasslands historically covered 162 million hectares ( ha ) of central and western North America (Samson and Knopf 1994) . These grasslands can be loosely subdivided based on characteristics of the dominant plant life forms into shortgrass, mixed grass, or tall grass prairies ( Coupland 1961, Samson and Knopf 1994, Lauenroth and Burke 2008; Fig . 1 ). In t his review we focus our attention on the shortgrass and northern mixed grass prairies of the Great Plains. The shortgrass prairie, or shortgrass steppe ecoregion, is dominated by C 4 grasses ( typically 6 ) including blue grama ( Bouteloua gracilis ), buffa lo grass ( Bouteloua dactyloide s ), and Western wheat grass ( Pascopyrum smithii ) and comprise s the s outhwestern portion of North American temperate grasslands ( Lauenroth et al. 1999, Lauenroth and Burke 2008; Fig . 1 ). Mixed grass prairies are split into two sub regions: n orthern mixed gr ass and s outhern mixed grass, extending n orth into Canada and s outh to n orthcentral Texas ( Lauenroth et al. 1999, Lauenroth and Burke 2008; Fig . 1 ). Mixed grass prairies are characterized by ecotones of shortgrass steppe and s agebrush shrubland, and support rich communities of forbs like goldenrod ( Solidago sp. ), slimflower scurfpea ( Psoralea tenuiflora ) , and scarlet globemallow ( Sphaeralcea coccinea ) (Lauenroth et al. 1999, Adler and HilleRisLambers 2008) . Common shrub species may include Wyoming bi g sagebrush ( Artemisia tridentata wyomingensis ) and greasewood ( Sarcobatus vermiculatus ) ( Lauenroth et al. 1999,

PAGE 11

3 Adler and HilleRisLambers 2008 ) . Collectively, the vegetation structure of these grasslands support diverse communities of interactive fauna (Samson and Knopf 1994, Augustine and Baker 2013, Grant et al. 2017) , presenting us with an opportunity to highlight potential trophic level cascades that inform future grassland management objectives. Figure 1. Map of North Ameri can temperate grassland categories and regions. Pawnee National Grassland represents a shortgrass prairie and Thunder Basin National Grassland represents a northern mixed grass prairie. Data retrieved from Esri REST Services Dir ectory.

PAGE 12

4 Focal Species Our hypothesis concerning trophic cascades in this system hinges on the role of black tailed prairie dogs in structuring requisite habitat for these species. The e cological benefits of black tailed prairie dogs are extensive: this s pecies engineer s resource ri ch environment s for grassland birds, small to medium sized carnivores, and many burrowing species by providing habitat for breeding, rearing of offspring, foraging, and cover from predators (Plumb et al. 2005, Hoogland 2006, Mar tin et al. 2007, Thiele et a l. 2013) . The US Fish and Wildlife Service (USFWS) reported a historically low 364,000 active h ectares of prairie dog colonies in 1961 (USFWS 2002, Virchow and Hygnstrom 2002) , and more recent estimates sugges t a range wide decline of 90 98% (Augustine and Baker 2013) , leading to their status of species of conservation concern (CPW 2015, US DA 2017, WGFD 2017) . These declines have also been linked to decreasing populations of swift f ox, burrowing owl, and mountain plover, three grassland species of conservat ion concern with the potential for trophic interactions (Desmond et al. 2000, Nicholson et al. 2006, Augustine et al. 2008, Alverson and Dinsmore 2014) . Swift F ox The s wift fox ( V ulpes velox ; Fig. 2 ) is a small predatory canid that frequently occup ies pra irie dog colonies and relies on denning opportunities and abundant prey resources therein (Kitchen et al. 1999, Kintigh and Anderson 2005) . Shrub free expanses across the shortgrass prairie like open fields and burned areas provide additional habitat for s wift fox (Thompson et al. 2008) . In regions characterized by mixed grass prairie and sagebrush steppe, swift fox may be specifically tied to prairie dog colonies because shrub composition and structure in these

PAGE 13

5 landscapes is low . Swift fox are a species of conservation concern at both state and federal levels (CPW 2015, US DA 2017, WGFD 2017) . Burrowing Owl Western burrowing owls ( Athene cun icularia hypug a ea ; Fig. 2 ) are small, fossorial raptors that lay eggs, rear chicks, and shelter in pre excavated burro w systems (Poulin et al. 2011) . Today, burrowing owls regularly colonize urban environments ( e.g ., golf courses, landfills, and storm drainage systems ) , remain abundant in agroecosystems, and frequently associate with both urban and rural prairie dog compl exes (Smith et al. 2005, Williford et al. 2009, Manning 2011, Mac í as Duarte and Conway 2015, Chi 2016) . In the Great Plains, o wls have been documented to respond to prairie dog alarm calls, providing evolutionary evidence that owls use prairie dogs for det ecting predators (Bryan and Wunder 2014) . Prairie dog colonies also support populations of ground nesting birds and insects; both prey items for burrowing owls. Burrowing owls remain a species of conservation concern at state and federal levels (CPW 2015, US DA 2017, WGFD 2017) , and North American Breeding Bird Survey (BBS) data between 1966 2011 estimate an annual population decline of 1 .1% (Sauer et al. 2013) . Mountain Plover Mountain plover ( Charadrius montanus ; Fig. 2 ) are upland shorebirds that nest, r ear young, and forage on landscapes with a relatively high bare ground structure , taking advantage tendency to clip vegetation and create short grass habitat (Augustine and Baker 2013) . Plovers do not exclusively depend on prairi e dog colonies for breeding habitat, as pastures moderately to heavily grazed by cattle, burn ed areas , and fallow crop fields throughout the s outhern portion of the species range provide favored habitat requirements

PAGE 14

6 (Wunder et al. 2003, Augustine and Skage n 2014, Ramsdell et al. 2016) . Prairie dog colonies; however, provide a greater source of habitat for breeding mountain plover s in the mixed grass prairies of the Northern Great Plains ( Wyoming, Montana, Canada ; Dinsmore et al. 2005, Plumb et al. 2005 ) . He re, shortgrass prairies transition into mixed grass/sagebrush steppe landscapes where prairie dogs often create the only shortgrass/bare ground habitat suitable for nesting . M ountain plovers remain a species of conservation concern throughout their range ( CPW 2015, US DA 2017, WGFD 2017) and North American Breeding Bird Survey (BBS) data between 1966 2011 estimate an annual p opulation decline of 3.0% (Sauer et al. 2013) . Figure 2. The swift fox ( Vulpes velox ; left ), western burrowing owl ( Athene cunicular ia hypugaea ; center ) and mountain plover ( Charadrius montanus ; right ) all directly benefit from black tailed prairie dogs ( Cynomys ludovicianus ) where their respective ranges overlap (Image credit: Cristi Painter). Understanding Trophic Ecology Managemen t objectives for prairie dogs should be geared towards meeting the requirements of multiple associated speci es of conservation concern. Reaching these objectives first requires a careful understanding of potential cascading trophic interactions between spe cies.

PAGE 15

7 Trophic Cascades Trophic cascade s can be classified as facilitating top down or bottom up responses (Elmhagen and Rushton 2007) . Top down control is driven by apex or dominant predators ( predators at the top of the food chain with no direct threat of predation ) or quaternary consumers ( species at the top level of the food chain capable of consuming all lower level species ; Lindeman 1942 ) triggering a response from lower level predators, consumers, or producers in the food chain ( Elmhagen and Rushton 2007, Ripple et al. 2016; Fig . 3 ). Bottom up control is characterized by primary producers o r lower level consumers driving changes to patterns in plant or wildlife populations and distributions higher up the food chain ( Elmhagen and Rushton 2007, Ripple et al. 2016; Fig . 3 ). Here, we focus on top down trophic cascades facilitat ing responses in p lant or wildlife communities that may in turn structure food chains and ecosystems (Paine 1980, Elmhagen and Rushton 2007, Ripple et al. 2016) . We do not discount; however, that bottom up mechanisms may be simultaneously at work and we support this by eval uating impacts to the vegetation community that may cause upward influences along the food chain. Responses may manifest as oscillations over time, as constant inc reases or decreases, or as static in populations across all trophic levels (Polis and Holt 19 92, Jiang et al. 2009) . These responses have repercussions for management, particularly when communities are comprised of multiple species of conservation concern (Soul é et al. 2005) .

PAGE 16

8 Figure 3. General processes of top down interactions ( left ) where upper trophic level species control populations of lower trophic level species, and bottom up interactions ( right ) where lower trophic level species control populations of upper trophic level species (Elmhagen and Rushton 2007, Ripple et al. 2016) . Intraguild Predation and Mesopredator Release Top down control occurs in many predator/prey relationships where intraguild predation and mesopredator release is present. We define intraguild predation as predators across different trophic levels compet ing for a shared prey resource, while the lower trophic level predator s also risk predation from the upper trophic level predator s (Polis and Holt 1992) . This process fits wi thin a broader, widel y accepted predator/prey hypothesis commonly referred to as m esopredator release . Here, apex predators become rare or absent in an ecosystem resulting in population spikes of smaller, medium level (meso) predators (Crooks and Soul é 199 9, Prugh et al. 2009) . M esopredators exploit prey resources at little cost of predation or competition from apex predators and initiat e top down control on lower trophic level prey communities ( Crooks and Soul é 1999, Elmhagen and Rushton 2007, Prugh et al. 2009, Ripple et al. 2016; Fig. 3 ). To

PAGE 17

9 further explore these hypotheses, we introduce a case study focusing on the dynamics between swift fox, burrowing owl, and mountain plover in the grasslands of the Great Plains. C ase S tudy Trophic Interactions on th e Pawnee and Thunder Basin National Grasslands To examine potential cascading trophic interactions within grassland systems we explored raw data trends (Fig. 4) in a nnual prairie dog colony mapping ( i.e. total colony boundary; surface area measured in hect ares (ha) ) an d count data for swift fox, burrowing owl, and mountain plover for two federally managed U . S . Forest Service (USFS) National Grasslands; (1) the Pawnee National Grassland (PNG) in Weld County of northeastern Colorado representing a shortgrass prairie ( USFS 2016a; Fig. 1 ), and (2) the Thunder Basin National Grassland (TBNG) in Converse, Weston, and Campbell Counties of eastern Wyoming representing a n orthern mixed grass prairie ( USFS 2016b; Fig. 1 ). Data Collection Both data sets are used annu ally by the USFS to inform manage ment of viable populations of prairie dogs and associated species (USFS 2016a, USFS 2016b) , and are collected under a collaborative effort by university researchers, federal wildlife managers , an d trained volunteers . Total prairie dog colony hectares (ha) on both USFS grasslands are attained by driving the outer boundary of active colonies and mapping those boundaries with GPS (Global Positioning System) units between June and September. Active co lonies are determined by the presence of prairie dogs, to include: fresh scat, freshly clipped vegetation, fresh digging at burrow entrances, and visual/aural identification. Only burrows that indicate evidence of active

PAGE 18

10 prairie dogs are included in the ma pping effort. Mountain plove r and burrowing owl surveys occur in early morning and late evening by driving transects no more than 400m apart across prairie dog colonies, stopping every 30 60 seconds to scan for the presence of adult birds during the breedi ng season (mid May through m id July). Swift fox surveys are conducted mid August through September via spotlighting at night while driving transects across prairie dog colonies or along designated USFS grassland roads adjacent to prairie dog colonies to de tect and confirm presence by eye shine. Surveys are conducted one time per colony, and total observations per colony are included into the annual data set. All survey methods follow USFS survey protocol (USFS 2016a, USFS 2016b) . Count data belongs to the U S Forest Service and will be made available upon request. Figure 4. Annual count data in raw form collected on (a.) the Pawnee National Grassland (PNG) from 1998 2016 and (b.) the Thunder Basin National grassland (TBNG) from 2010 2016 for total active prairie dog colonies me asured in hectares (ha) and population counts for swift fox, burrowing owl, and mountain plover (USFS 2016a, USFS 2016b) .

PAGE 19

11 Th ese raw data (Fig. 4) were collected under protocol s targeted for agency manage ment purposes , and not designed to address researc h objectives (USFS 2016a, USFS 2016b) ; therefore, inferences concerning processes in trophic ecology are purely speculative. We fit an exponential curve to the data (Fig. 5) to heuristically observe pattern s suggesting that the total area of active prairie dog colonies ha s increased across the PNG, with number of observed burrowing owls increasing and swift foxes and mountain plovers remaining uncommon and moderately declining (1998 2016; Fig . 5 a). Total are a of active prairie dog colonies ha s also increase d across TBNG; however, number of observed swift fox and mountain plover are increasing and burrowing owls are only responding weakly (2010 2016; Fig . 5 b). Combining these observed co occurrence patterns wi th examples of trophic ecology defined in the lite rature suggest the possibility of a cascading trophic interaction in this system , where the presence or absence of swift fox directly influences the abundance of mountain plovers by controlling burrowing ow l mesopredators . Factors beyond increased prairie dog colony area are likely at play because species responses differ markedly between the PNG and TBNG during the same period.

PAGE 20

12 Figure 5. An exponential curve fitted to annual trend data collected on (a.) the Pawnee National Grassland (PNG) from 1998 201 6 and (b.) the Thunder Basin National grassland (TBNG) from 2010 2016 for total active prairie dog colonies measured in hectares (ha) and population counts for swift fox, burrowing owl, and mountain plover (USFS 2016a, USFS 2016b) . Hypotheses Concerning Trophic Ecology on Pawnee and Thunder Basin National Grasslands Mesopredator Release on the PNG M esopredator release may explain the increase of burrowing owls on the PNG and their co occ urrence patterns with swift fox and mountain plover . Bu rrowing owls may respond positively to declining swift fox populations (Fig . 5 a) pressure of a higher level predator . Hunting and trapping presents an annual additive mortality

PAGE 21

13 factor for swift fox in Colorado with an estimated har vest of over 600 individuals in the 201 4 201 5 season (CPW 2016, Stukel 2017) ; however, uncertainty around how to obtain accurate harvest estimates of swift fox remains a ch allenge for state wildlife agencies (CPW 2016) . The current population of swift fox in eastern Colorado is unclear; however, state managed occupancy surveys occurring every 5 years suggests that eastern Colorado still supports the largest population of swi ft fox throughout their range (Stratman 2017) . Predation from coyotes is also a driv er of swift fox mortality in Eastern Colorado (Kitchen et al. 1999, Karki et al. 2007) . Kitchen et al. found that coyotes contributed to 48% of swift fox mortality in a res ource partitioning study in s outheastern Colorado (Kitchen et al. 1999) . Complex tro phic interactions between canid sp. are documented in the literature (Robinson et al. 2014, Lonsinger et al. 2017) ; however, the extent to which coyotes currently impact sw ift fox on the PNG remains uncertain. Because coyotes have been documented to influe nce swift fox abundance (Kitchen et al. 1999, Karki et al. 2007) , coyote presence should be considered as a driving mechanism in future trophic research between foxes, owls , and plovers (Fig. 6) on black tailed prairie dog colonies on the PNG.

PAGE 22

14 Figure 6. (a.) Swift Fox ( Vulpes velox ), (b.) Western Burrowing Owl ( Athene cunicularia hypugaea ), and (c.) Mountain Plover ( Charadrius montanus ) are all associated species of (d.) Black tailed Prairie Dog ( Cynomys ludovicianus ) colonies. We hypothesize that foxe s consume owls (solid arrow) and owls consume plovers (solid arrow). Foxes also consume plovers (dashed arrow) and prairie dogs (dotted a rrow), making this a highly interactive predator/prey community. (e.) Coyotes ( Canis latrans ) are also present in the s ystem and play an unknown role in the hypothesized cascade by potentially influencing swift fox abundance, and preying opportunistically on all species (Image credit: Cristi Painter, Ryan A. Parker, and © Can Stock Photo Inc.). Regardless of the cause, a decline in the count of swift fox on the PNG is concurrent with the increased count of burrowing owls and decreased count of mo untain plover ( USFS 2016a; Fig. 5a ). Release of owl mesopredators may be linked to increasing infrequency of mountain plovers on prairie dog colonies on the PNG despite expansion of suitable habitat. Lower trophic level prey resource s often decline after t he eruption of a mesopredator due to increased predation (Crooks and Soul é 1999, Prugh et al. 2009) . B urrowing owls consume moun tain plover nestlings (Conrey 2010) support ing th is hypothesized trophic cascade between foxes, owls, and plovers (Fig. 6). We also note that radio transmitters attached to plovers have been recovered inside swift

PAGE 23

15 fox dens, suggesting that plovers are a sh ared prey resource for both foxes and owls (Mille r and Knopf 1993, Conrey 2010) . Under this scenario, an increased predation effect on plovers by owls may outcompete a predation effect on plovers by foxes. Mesopredator Release on the TBNG P redator/prey dynamics consistent with intraguild pred ation (Polis and Holt 1992) compare with co occurrence patterns see n on TBNG. Swift fox remain common and protected from hunting and trapping in the n orthern mixed grass prairies of e astern Wyomin g ( USFS 2016b, WGFD 2017 ) , and spotlight surveys suggest tha t this region remains a stronghold for swift fox. Although burrowing owl presence indicates a moderate increase on TBNG, trends remain low despite increasing availab ility of suitable habitat ( USFS 2016b; Fig. 5b ). Mo untain plover populations mirror the trend of swift fox, suggesting that even a minor increase in the count of burrowing owl s still potentially alleviates predation o n plovers which allows them to take advantage of expanding prairie dog colonies, refl ectin g patterns of intraguild predation between owls and foxes ( USFS 2016b; Fig. 5b ). Anecdotally, burrowing owls are selecting for resource deficient habitats where swift fox are less likely to persist ; t he shared prey resource, mountain plover, is respon ding positively to the owls change in habitat selection and the increased acreage of prairie dog colonies . Coyotes are also present on the landscape on TBNG; however, the degree of impact on lower trophic level canid species remains undocumented. Coyote popu lations are lethally controlled in eastern Wyoming by local landowners, hunters, and predator control groups and are infrequently documented on prairie dog colonies during associated species abundance surveys

PAGE 24

16 (USFS 2016b) . Successfully controlled coyo te po pulations on the TBNG provides support for an increased abundance of swift fox; however, this inference remains speculative based on conversations with USFS biologists (USFS 2016b) . Because coyotes have been documented to influence swift fox abundance in C olorado (Kitchen et al. 1999, Karki et al. 2007) , coyote presence should be considered as a driving mechanism in future trophic research between foxes, owls, and plovers (Fig. 6) on the TBNG in Wyoming. Landscape F ragmentation D riving M esopredator R elea se Conversion of grassland habitats may be an important driver of predator/prey dynamics because cascading t rophic interactions have been linked to landscape fragmentation (Crooks and Soul é 1999, Prugh et al. 2009) . Prugh et al. (2009) point to three fact ors that lead to declines in populations of top level predators in fragmented landscapes: (1) Apex predators have expansive territories and need large, connected areas of in tact habitat (Prugh et al. 2 009) . Therefore, as landscapes become fragmented, apex predators are less likely to meet resource requirements and become uncommon or absent from the area. (2) Fragmented landscapes provide attractive habitat for mesopredators because apex predators become rare and food resources in the form of human waste or crop fields become plentiful (Prugh et al. 2009). And (3), in fragmented landscapes, apex predators experience frequent human conflict leading to lethal predator control (Prugh et al. 2009) . Thus, w e uphold that predator driven dynamics and the degree of habitat fragmentation could also support our hypot hesis that swift fox, burrowing owl, and mountain plover are involved in a complex trophic interaction on black tailed prairie dog colonies (Fig . 6 ) . Landscape F ragmentation on the PNG

PAGE 25

17 Shortgrass prairie on the PNG is intermixed with independently manage d rangeland and privately owned agricultural cropland, the latter fragments the landscape by transforming native grassland into crop land . Here, individual prairie dog colonies remain small (~ 0.5 138 ha, 201 6 mapping data ) and dispersed (USFS 2016a) . Spatia l and temporal changes to prairie dog colonies partly result from agricultural development, but also from Yersinia pestis , a n infectious bacterium that causes sylvatic plague ( a disease often lethal to prai rie dogs ; Eads and Biggins 2017 ). Plague is present annually on the PNG, clearing out over 95% of the prairie dogs in any given colony (Webb et al. 2006, Eads and Biggins 2017) . Annual prairie dog mapping reveals periodic appearance and disappearance of in dividual active colonies that shif t geograph ically across the landscape between years (USFS 2016a) . Thus, a combination of agricultural development and plague contributes to the fragmentation and small size of prairie dog colonie s . T he total number of sm all, isolated colonies are rising on the PNG (USFS 2016a) ; however, s uggest ing that benefits to swift fox from prairie dogs may be negated by the surrounding landscape fragmentation. The transformation of grassland into agricultural cropland has removed ex panses of swift fox habitat (Kamler et al. 2 003) , and small colony sizes combined with annual geographical changes in colony placement may further i nfluence swift fox presence. These impacts may describe the increasing trend of burrowing owls ( USFS 2016a; Fig. 5a ). Given that owls cue in on prairie dogs (Wunder and Bryan 2014) , geographical changes in colony placement may not impact owl presence provid ed prairie dogs remain on the landscape. Additionally, owls will continue to breed on colonies impacted by plague providing burrows remain intact , vegetation does reach excessive heights, and prairie dogs re colonize shortly after (Butts and Lewis 1982, Conrey 2010, Alverson and Dinsmore 2014) . T his suggests that

PAGE 26

18 fragmented prairie dog colonies on the PNG have little impact on burrowing owls , further supporting the mesopredator release hypothesis (Crooks and Soul é 1999, Prugh et al. 2009) . Considering these fragmentation effects, mountain plover d eclines on prairie dog colonies may be respon ses to top down contr ol due to increased predation and/or selecti on of other suitable nesting habitat in the vicinity such as prescribed burns and fallow crop field s . Regardless , the conversion of shortgrass pra irie to agricultural cropland potentially benefits mountain plover s and alludes to a case of landscape fragmentation that both negatively and positively impacts different levels of the potential trophic cascade . Landscape Fragmentation on t he TBNG Landscape fragmentation impacts the n orthern mixed grass prairie to a lesser extent. The soil structure is unsuitable for producing crops leaving cattle ranching as the primary agricultural practice (Fitzgerald et al. 1999) . On the TBNG, black tailed prairie dog colonies comprise large complexes upwards of ~12,000 ha span ning public and privately owned land. Annual mapping efforts dating to 1999 indicate that these expanses oscillate over time (Fig. 4) , declining when epizootics of plague sweep through the landscape (USFS 2016b) . Plague is less prevalent on an annual basis on TBNG and instead reflects patterns of density dependence (Cully et al. 2010) . Prairie dog colonies were recovering from a plague epizootic at the start of the trend da ta in 2010 ( USFS 2016b; Fig. 5b ), so using plague as a driver of fragmentation on TBN G helps address fox, owl, and plover interactions. Prairie dog colonies expanded across the grassland post 2010 ( USFS 2016b; Fig. 5b ), creating large, intact h abitat patches for associated species. Systems with intraguild predation are frequently docume nted occurring on intact habitat patches (Robinson et al. 2014, Lonsinger et al.

PAGE 27

19 2017) , providing an explanation for the relationship between co occurring foxes a nd owls. Foxes on the TBNG are found almost exclusively on prairie dog colonies and presence h as increased following the prairie dog expansion (USFS 2016b) . Burrowing owls are sparsely distributed across the prairie dog complexes and remain uncommon relati ve to the available habitat. If swift fox and burrowing owls are indicative of intraguild pred ation on TBNG, patterns in mountain plover trends follow suit : as fox occurrence increases, owls only minorly increase, and plovers increase (Fig . 5 b). Plovers st ill risk predation by swift fox and other grassland predators; however, predation risk from ow ls is reduced . Thus, these potential cascading trophic interactions may be facilitated, in part, by the availability of intact patches of habitat created by prair ie dogs. Bottom up C ontrol We have primarily focused on top down control as a leading hypot hesis to explain the trophic ecology behind swift fox, burrowing owl, and mountain plover interactions on black tailed prairie dog colonies. However, we cannot discount alternative ecological processes at work. Bottom up control is characterized by primary producers or lower level consumers driving changes to patterns in plant or wildlife populations and distributions at higher levels in the food web ( Elmhagen and Rushton 2007, Ripple et al. 20 16; Fig . 3 ). Both the shortgrass and mixed grass prairie communi ties are subject to changes in climate which change plant composition and structure (Polsky and Easterling III 2001) . Here, an increase in percent precipitation may lead to taller grasses and forbs across prairie dog colonies changing the habitat condition s required by associated species. Vegetation structure on a colony is also influenced by plague, when prairie dogs are no longer present to clip the vegetation back to a shortgrass, bare groun d state (Augustine et al. 2008) . Changes to plant composition an d structure lead to changes in the insect

PAGE 28

20 community (Whicker and Detling 1988) , or may represent an absence of prairie dogs, which may then cascade up the food web impacting species from the b ottom up. We argue that it is likely never the case where only t op down or bottom up control is at work, but instead both trophic level processes are working concurrently to influence the biotic state of the community. Therefore, carefully considering all factors that may contribute to a bottom up effect on a swift fox burrowing owl mountain plover interaction is critical for understanding all dynamics of the system. Additional I nfluences from USFS G rassland M a na gement Techniques The PNG and TBNG manage for prairie dogs, and to varying degrees, associated species. Prairie dog management consists of both lethal and non lethal management strategies. Lethal strategies on the PNG include non regulated prairie dog shooting and poisoning of colonies that interfere with agricultural operations . TBNG has regulated a prairi e dog shooting closure on colonies that are considered core areas for associated species. Local stakeholders successfully removed the shooting closure in recent history an d poisoning of prairie dogs on the grassland has increased. Non lethal prairie dog ma nagement includes structural ( i.e. vegetation/fence ) barriers on colon y edges to prevent dispersal in any given direction on both grasslands, and translocation of prairie dogs from unsuitable areas to patches of core area habitat on the TBNG. Plague miti gation in the form of application of an insecticide, d elta m ethrin, that targets fleas ( order Siphonaptera the vector of sylvatic plague ; Eads and Biggins 2017 ) on prairi e dogs has occurred on both the PNG and TBNG. Prescribed burns that create temporary habitat for m ountain p lover and swift fox have occurred on both USFS grasslands. Finally, grazing of domestic cattle occurs on pastures that overlap prairie dog colonies on both USFS grasslands. In areas where heavy grazing occurs outside of prairie dog co lonies, bare ground habitat can result ,

PAGE 29

21 which may attract dispersing prairie dogs and breeding plovers. Grazing periods, stocking rates, and densities may have an impact on the likelihood of prairie dog and associated species occurrence and all yield varyi ng levels of success. Thus , it is likely that management actions infl uence cascading tropic interactions by altering habitat composition and co occurrence between species . Conclusions Trophic cascades occur in a diverse range of ecosystems and exi st between flora and fauna at all trophic levels (Paine 1980, Elmhagen and Rushton 2007, Ripple et al. 2016) . Cascading t rophic interactions in North American temperate grasslands remain poorly described despi te a heavy management focus on many individual species in the system. Th e theoretical framework presented here provides a starting point for research on species interactions between swift fox, burrowing owl, and mountain plover on prairie dog dominated ran geland communities. We add that there are many alternative, plausible hypotheses that we do not describe in detail in this manuscript. For example, mountain plover populations on both grasslands may be largely driven by nesting habitat availability, to inc lude insect availability, and populations may r eflect these factors more so than direct or indirect influences from co occurring species. And burrowing owl populations may increase with a direct correlation from nesting habitat availability independent fro m a decrease in swift fox populations. We also acknowledge that it remains unclear which life stage predation on owls and/or plovers predominantly occurs. We fully acknowledge that i nferences made in this paper are anecdotal and maintain that future investigation into this system is needed. Trophi c cascades are a driving force in community ecology that can not be defined by any one mechanism. The presence or absence of

PAGE 30

22 top level consumers, hunting, disease, management strategies, agricultural p roduction, habitat fragmentation, or deviations in resou rce availability, all which change temporally, contribute to the dynamics of trophic cascades . Other e nvironmental variables not discussed here, such as climate, should also be considered. We propose t arget ed long term multi species occupancy research that combines the hypothesized mechanisms specified above with presence and co occurrence of associated grassland species. This level of analysis is recommended for exploring these complex interactions moving forward , and this manuscript only serves as a prelu de to this suggestion . Relationships between prairie dogs and associated species are highly interactive, so managing prairie dog colonies for viable populations of associ ated grassland wildlife is recommended from a conservation perspective . This promotes species richness , rangeland health, and diversity in shortgrass and mixed grass prairies, and highlights opportunities for investigating multi species interactions across all trophic level s .

PAGE 31

23 C HAPTER II FACTORS INFLUENCING THE OCCURRENCE OF SWIFT FOX, BUR ROWING OWL, AND MOUNTAIN PLOVER ON BLACK TAILED PRAIRIE DOG COLONIES IN WYOMING I ntroduction Many species of conservation concern depend on patches of habitat w ithin a broader landscape and fulfill a role in the ecological processes that drive their popul ations within these patches. The extent to which a given species can be associated with a habitat patch is influenced by variations in patch characteristics such as patch size, resource availability, human presence, disease, vegetation structure and divers ity, and climate ( Ye et al. 2013, Zuckerberg et al. 2018 ). Changes in these characteristics over space and time often drive habitat quality, which in turn may govern species occupancy on a given habitat patch. Inter specific interactions may also lead to c hanges in occupancy rates on a patch for individual species (Grinde and Niemi 2016), suggesting that occupancy is not only a function of habitat quality, but a lso a function of the presence of co occurring species. To quantify the effect of habitat quality and co occurring species presence on occupancy, we can evaluate influencing characteristics at the patch level. Patch characteristics can be measured alongsid e repeated presence or absence surveys for one or more species to estimate occupancy probabilitie s that inform applied management decisions (Estevo et al. 2017). Management decisions directed towards meeting the habitat requirements for multiple co occurri ng species work to mitigate declining animal populations (Sauer et al. 2013, White et al. 2013). Therefore, understanding the ecology behind population dynamics in a patchy environment is important for assessing metapopulations of wild animals, particularl y co occurring species of conservation concern.

PAGE 32

24 Black tailed prairie dogs ( Cynomys ludovicianus ) are a species of conservation concern at state and federal levels, and their colonies serve as individual patches of habitat in the Northern Great Plains for some species (Hoogland 2006, USDA 2017, WGFD 2017). Prairie dogs are small, social mammals that excavate complex tunnel systems that provide shelter for adults and offspring and a place to store food (Hoogland 2006). Above ground, prairie dog colonies are characterized by mounds of soil and clipped vegetation leaving a short grass, bare ground landsc ape that allows for predator detection (Hoogland 2006). Prairie dogs are considered a keystone species that provide habitat for over 100 grassland associated s pecies, and they serve a controversial role in rangeland and agricultural production by creating a bare ground landscape that is undesirable for livestock producers who rely on grasses and forbs for cattle forage (Hoogland 2006, Field et al. 2016). Prairie dog colonies change spatially and temporally across the landscape in response to habitat fragmen tation, disease, lethal and non lethal management, and climate, all of which contribute to changes in individual colony size, vegetation structure, and probabi lity of occupancy by colony associated species (Hoogland 2006, Johnson et al. 2011, Dinsmore and Smith 2010, Alverson and Dinsmore 2014). Collectively, the patch characteristics of a prairie dog colony support a complex system of ecological interactions. dynamic s of ecological interactions on a colony. For example, prairie dog colonies in the Northern Great Plains typically occur in mixed grass prairie mosaics where a variety of habitat types intermix (Duchardt et al. 2018, Parker et al. 2019). Here, species that strongly benefit from prairie dogs are almost exclusively tied to colonies because they provide the only source of suitable habitat in the region (Parker et a l. 2019). In this study, we address potential species interactions between three prairie dog asso ciates: swift fox ( Vulpes velox ), burrowing owl

PAGE 33

25 ( Athene cunicularia hypugaea ), and mountain plover ( Charadrius montanus ) that strongly benefit from prairie dog colonies in the Northern Great Plains (Parker et al. 2019). Foxes, owls, and plovers are also li sted as species of conservation concern at state and federal levels (USDA 2017, WGFD 2017). Swift Fox: The swift fox is a small canid that frequently occupies shrub free expanses across the Great Plains such as open fields, burned areas, and prairie dog c olonies ( Dark Smiley and Keinath 2003, Kintigh and Anderson 2005). Prairie dog colonies provide denning opportunities and abundant prey resources for swift fox es to include prairie dogs, rabbits, mice, and ground nesting birds (Miller and Knopf 1993, Kinti gh and Anderson 2005). Burrowing Owl : Burrowing owls are small raptors that nest and shelter in pre excavated burrows (Poulin et al. 2011). In the Great Plains , owls have been documented on prairie dog colonies to respond to prairie dog alarm calls, sugges ting that owls use prairie dogs for detecting predators (Bryan and Wunder 2014). Prairie dog colonies support abundant populations of ground nesting birds and insects; both prey items for burrowing owls (Conrey 2010). Mountain Plover: Mountain plovers are upland shorebirds that nest and forage on landscapes with relatively high bare ground; a characteristic of prairie dog colonies (Augustine and Baker 2013). Pra irie dog colonies are a strong source of habitat for breeding mountain plovers in the Northern Gr eat Plains, where shortgrass prairies transition into mixed grass/sagebrush steppe ecotones (Duchardt et al. 2018). Parker et al. (2019) proposed that these s pecies are involved in a complex web of predator and prey interactions that reflect patterns rese mbling a trophic cascade. Because prairie dog colonies change spatially and temporally in the Northern Great Plains, and because foxes, owls, and plovers are s trongly associated to these colonies (Dinsmore and Smith 2010, Alverson and Dinsmore 2014, Duchar dt et al. 2018), we investigated whether the probability of occupancy

PAGE 34

26 of a colony by a given species is influenced by the presence of co occurring species at d ifferent trophic levels, as well as colony size. Using individual prairie dog colonies as our sam ple unit, we used a hierarchical modeling approach (Royle and Kéry 2007, Hobbs and Hooten 2015) to estimate single species occupancy probabilities of prairie d og colonies by foxes, owls, and plovers under the following hypotheses (Box 1). M ethods Study area We studied black tailed prairie dogs, swift fox, burrowing owl, and mountain plover from 2016 2018 on the Thunder Basin National G rassland (TBNG) in the mixed grass prairies of northeast Wyoming (Figure 1). The ~1600 km 2 study area is bordered by Wyo ming State Highways 450 to the north and 50 to the west, the Miller Hills to the south, and contiguous private land to the east. TBNG i s comprised of a mixture of state owned or federally managed Box 1. Hypot heses regarding occupancy of black tailed prairie dog colonies by swift fox, burrowing owl, and mountain plover: 1) Swift Fox: a. Occupancy is informed by the combined presence/absence of Burrowing Owls and Mountain Plovers. b. Occupancy is informed by prairie dog colony size. 2) Burrowing Owl: a. Occupancy is informed by the combined presence/absence of Swift Fox and Mountain Plovers. b. Occupancy is informed by prairie dog colony size. 3) Mountain Plover: a. Occupancy is informed by the combined presence/absence of Swift Fox and Burrowing Owls. b. Occupancy is informed by prairie dog colony size.

PAGE 35

27 US Forest Service (USFS) grassland and privately owned ranch es. Due to access limitations, this research was conducted on public lands only. TBNG is characterized as a northern mixed grass prairie, consisting of ecotones of sage brush ( Artemisia sp. ) steppe, ponderosa pine ( Pinus ponderosa ) and Rocky Mountain junip er ( Juniper us scopulorum ) forests, riverbeds lined with plains cottonwoods ( Populus deltoides ), rocky outcroppings and escarpments, and patches of short grass, bare ground prairie typically occupied by prairie dogs (USDA 2001, Duchardt et al. 2018). Grasse s and forbs comprise the vegetation structure on prairie dog colonies, to include western wheat grass ( Pascopyrum smithii ), blue grama ( Bouteloua gracilis ), prairie junegrass ( Koeleria macrantha ), plains prickly pear ( Opuntia polyacantha ), scarlet globemal low ( Sphaer alcea coccinea ), wooly plantain ( Plantago patagonica ), and pepperweed. ( Lepidium sp. ). Figure 1. Thunder Basin National Grassland (TBNG) is located in northeastern Wyoming. Sampled black tailed prairie dog ( Cynomys ludovicianus ) colonies, i n black, ra nged in size from 2 ha to 2767 ha with a mean of 456.64 ha and median of 146.5 ha ( see Table 2b) . Map created in Esri ArcMap 10.4.1.

PAGE 36

28 Prairie dog colonies (patch sample unit) Within the study area, we surveyed foxes, owls, and plovers exclusivel y on black tailed prairie dog colonies (Figure 1). We used prairie dog colonies as our sample unit because each colony is a stand alone habitat patch that requires a unique survey effort . Each colony was y ear survey. All colonies occurred within TBNG boundaries and were all previously identified and mapped by USFS biologists or contracting biologists. Prairie dog colonies have been mapped in this landscape annually since 1999 to record spatial and temporal variation in colonies between years. Colonies were mapped by Trimble or Garmin GPS units to mark the physical outermost boundary of active prairie dogs identified by the sight and sound of prairie dogs, evidence of active digging and fresh scat, and in tac t burrow systems. Burrows on the outermost boundary that were collapsed or sealed off suggest prairie dog absence or inactivity and were excluded from mapping. On TBNG, prairie dog colon ies are typically separated by stands of woody vegetation or deep drai nages and gullies, and Ulev (2007) reports that burrows in the Northern Great Plains typically occur on <10% slope. Field observations and colony mapping aligned with this report and ind icated that steep hillsides and rolling hills also divided colony boun daries on TBNG. Because prairie dog colonies on TBNG are annually susceptible to sylvatic plague (Cully Jr. et al. 2010, Eads et al. 2018), lethal and non lethal management, and changes in climate, mapping efforts tracked temporal changes in colony charact eristics. All mapping shapefiles used in this study were retrieved from USFS biologists or contracting biologists and were analyzed in ArcMap 10.4.1 to extract colony area for all coloni es surveyed in each of our sample seasons. Multi species sampling

PAGE 37

29 We recorded the presence or absence of swift fox, burrowing owl, and mountain plover from 2016 2018 within a breeding period survey window of May 1 st and July 15 th each year. Because breedi ng periods for all three species overlap and rearing of offspring carr ies into late July, we combined survey efforts for all three species into one multi species survey and looked for all three species simultaneously. To account for occasions when species went undetected on a colony during a survey but remained present, we c onducted repeat surveys on each colony three times within the survey window (MacKenzie et al. 2006). Surveys occurred from sunrise to ~11:00 a.m. and ~5:00 p.m. to sundown to avoid the h ighest temperatures of the day. Surveys included driving weaving trans ects ( i.e. zig zag motion ) across prairie dog colonies no more than 400m apart by truck or ATV, stopping every time a detection was made, or every ~15 seconds to scan. Driving in weaving transects aided in flushing nesting plovers and provided a more exhau stive survey across the habitat patch for all species. Our research rests on the assumption that the probability of occupancy remains the same at individual prairie dog colonies within a season and across all years for each species. A violation to this ass umption occurs in the event of nest predation or when increased human presence results in mortality or the relocation of an adult individual to a new habitat patch. We assumed that prair ie dog colonies are independent and that detecting a species at one si te is independent of detecting the same species at a different site (Bailey and Adams 2005, MacKenzie et al. 2006). We monitored all detected species to identify whether an active nest o r den existed at a site and then matched adult individual(s) to a nest or den on the colony being sampled. Finally, we use colony size and associated species presence/absence as individual site covariates to quantify differences between probabilities of oc cupancy across all sampled colonies.

PAGE 38

30 Single species hi erarchical framework We created single species, robust design occupancy models (MacKenzie et al. 2006) under a Bayesian hierarchical framework using JAGS (Package R2jags, R version 3.5.1 , Royle and Kéry 2007, Hobbs and Hooten 2015, Su and Yajima 2015) to i nvestigate the impacts of associated species presence and colony size on the probability of occupancy of prairie dog colonies by foxes, owls, and plovers. We created a set of models for each species that pools across all years and treats the three individu al sampling seasons as a single season survey. We pool across all years to increase sampling size instead of stratifying by individual year because our sparse data set prevents model con vergence when analyzing between sampling seasons ( see Appendix A for m odels stratified by year). We acknowledge that pseudoreplication (Hurlbert 1984) occurs under this approach; however, we continued with analysis to determine whether covariate effects co uld be identified on occupancy probability estimates with an increased sample size. For all models, we generated prior distributions for Beta (distribution of covariates) and the probability of detection ( P ), using the logistic and uniform distributions, r espectively (Northrup and Gerber 2018, Outhwaite et al. 2018). The fol lowing likelihood framework was used to model the logit probability of occupancy ( ): Likelihood: logit( [i]) = Beta 0 + Beta 1 * X1[i] + Beta Beta N * XN[i] Z[i] = bernouli( [i]) Y_Sp[i] = binomi al((Z[i] * P ), J[i]) Equation 1. Here, X represents a patch level covariate (colony size or co occurring species presence/absence), i is the individual habit at patch (colony), Z represents the true state of occupancy, Y_Sp refers to the to tal number of detections of a single species ( Y_SwiftFox,

PAGE 39

31 Y_BurrowingOwl, Y_MountainPlover ), and J is the total number of repeat visits in a season to habitat patch i (Equati on 1). These models yield an estimate for the probability distributions of paramet ers psi ( ) and P , occupancy and detection, respectively. These parameters inform the proportion of prairie dog colonies occupied by each species, and the probability of dete cting each species 1 or more times on a given colony. To further refine our estima te for , we included a set of covariates as priori predictions and kept P constant. Because we are modeling sparse data, holding P constant lessens the number of parameters estimated and produces a more reliable result for (Welsh et al. 2013), our param eter of interest. Prairie dog colonies change spatially and temporally between years impacting the ecological characteristics of the sample unit and the species that are present. However, our data set only covered a three year span (2016 2018) with small sample sizes in each year, so we pooled across years and assumed closure between all years. We included year as a covariate in a separate set of models and include the results and the list of non converging models for this set in the appendices (Appendix A ). To further refine our estimates of , we evaluated covariate effects from associated species presence and colony size. Associated species presence: We surveyed for swift fox, burrowing owl, and mountain plover simultaneously on prairie dog colonies and each species received an individ u al presence score, 1, or an absence score, 0, based on the detection of that species during the survey. When estimating swift fox occupancy, we used the combined three survey presence and absence scores for burrowing owl on each colony plus the presence a n d absence scores for mountain plover on each colony as a patch site covariate (Table 1). When estimating burrowing owl occupancy, we used the combined three survey swift fox presence/absence scores plus mountain plover

PAGE 40

32 presence/absence scores as a patch s i te covariate (Table 1). Finally, for estimating mountain plover occupancy, we used the combined three survey presence/absence scores for swift fox plus presence/absence scores for burrowing owl (Table 1; see Robinson et al. 2014 for a similar approach). T h ese covariates were used to address our initial hypotheses (Box 1) to explore proposed species interactions that occur on prairie dog colonies (Parker et al. 2019). We used the presence/absence covariates for two species to measure an interaction effect o n occupancy for the third species because we are interested in species co occurrence on any given colony (Table 1). Prairie dog colony size: Patch size is frequently included in ecological models to estimate influences on species occurrence because it is often informative for determining the proportion of habitat necessary for a given species to meet its resource needs (Martinson et al. 2012, Shake et al. 2012). For example, prairie dog colonies in Phillips County, Montana, show a quadratic effect o n colony size for probability estimates of occupancy of a colony by both burrowing owl and mountain plover (Dinsmore and Smith 2010, Alverson and Dinsmore 2014). This suggests that colony size plays an important role in the habitat selection process by an i ndividual. We use colony area in hectares (ha) from the previous year (2015 2017) for each year surveyed (2016 2018, pooled across all years ) as a covariate to estimate a patch size effect and a quadratic patch size effect on probability estimates of occu p ancy (Table 1). Using a previous year patch size approach is supported biologically because the success of a species on a given patch can predict the fidelity of that species to a habitat patch in the following breeding season (Schmidt 2001). Raw data for colony size recorded in hectares was scaled ((area mean) / standard deviation) for modeling in R2jags.

PAGE 41

33 Table 1. Exhaustive list of models created to estimate the effect of co occurring species presence and colony size on the probability that a pra i rie dog colony is occupied by swift fox, burrowing owl, and mountain plover. Here, letter a represents one of the three species, while letters b and c represent the combined three survey presence/absence scores for the two remaining species. R es ults We surveyed 24 colonies in 2016, 34 colonies in 2017, and 30 colonies in 2018 for a total sample size of 88 colonies over the three year period (Table 2a). We recorded a total presence of swift fox on 25 colonies, burrowing owls on 34 colonies, and mo untain plover on 42 colonies across all years (Table 2a). 36 of our sampled colonies were between 0 to 100 ha in size recorded from the previous year to sampling, representing 41% of the total colonies surveyed (Figure 2). 14 of our sampled colonies were > 1000 ha in size recorded from the previous year to sampling, representing 16% of the total colonies surveyed (Figure 2). Colony size across all years ranged from 2 ha to 2767 ha with a mean of 456.64 ha and a median of 146.5 ha (Table 2b). Table 2. a. ) N umber of colonies sampled each year, along with number of presence scores recorded for each species; b. ) Range, mean, and median colony size (ha) pooled across years.

PAGE 42

34 Figure 2. Distribution of n=88 black tailed prairie dog colonies by size (hectares) on TBNG, representing the previous year colony size (2015 2017) for each year sampled (2016 2018). Our models pool data across all sampling seasons to increase our sample size to n = 88 (Table 2a) and test an effect of colony size and associat ed species presence on occupancy (Table 1). Swift Fox: Bayesian trace plots retrieved from R2jags output (Appendix A ) and model estimates indicate that the best converging model in this set was the intercept model that holds occupancy and detection constan t ( = .649 ± .159, lower CI =.379, upper CI = .964 ; Table 3a).

PAGE 43

35 We were unable to reliably estimate occupancy for the remaining models ( is near or at 1.000; Table 3a). Burrowing Owl: Trace plots and model estimates indicate strong model convergence for t he intercep t model ( = .475 ± .070, lower CI = .348, upper CI = .623 ; Table 3b) and the model testing a linear effect on colony size ( = .479 ± .078, lower CI = .341, upper CI = .644 ; Table 3b), and poor model convergence for all other models. Mountain P lover: Trac e plots and model estimates also indicate strong model convergence for the intercept model ( = .604 ± .073, lower CI = .467, upper CI = .755 ; Table 3c) and the model testing a linear effect on colony size ( = .680 ± .101, lower CI = .500, uppe r CI = .897 ; Table 3c), and poor model convergence for all other models. In all three model sets, certain models produced probability estimates for occupancy that were smaller than the lower CI (Table 3a c); a further indication of sparse data. Table 3. Analyzed mo del sets for a. Swift Fox, b. Burrowing Owl, and c. Mountain Plover. Covariates include colony size (area), a quadratic effect on colony size (area 2 ), and the presence of swift fox (sf), burrowing owl (bo), and mountain plover (mp) and an inter action effec t (*) between two of the species to estimate a probability of occupancy ( ). Standard deviation and lower and upper 95% credible intervals are provided for the mean probability estimates of .

PAGE 44

36 We further evaluated the effect of colony size o n the odds t hat a prairie dog colony is occupied by burrowing owls and mountain plovers by analyzing the coefficient estimates of our converging models. The mean logit Beta coefficient estimate for colony size on the probability of occupancy of a prairie d og colony by burrowing owl is .339 ± .306. We converted this estimate to odds (e 2.033 ± 1.049 ) and report that for every 1 unit ha increase in colony size, the odds that a prairie dog colony is occupied by a burrowing owl increases by 1.403 ± 1.358. The me an logit Bet a coefficient estimate for colony size on the probability of occupancy of a prairie dog colony by mountain plover is 2.033 ± 1.049. Thus, for every 1 unit ha increase in colony size, the odds that a mountain plover occupied a prairie dog colony increased b y 7.637 ± 2.854. D iscussion We are unable to report on the effect of associated species presence or colony size on the probability that a prairie dog colony is occupied by a swift fox. We had n=25 swift fox presence records over a pooled sampling size of n=88 prairie dog colonies (Table 2a), which resulted in the failure of our models to properly converge. We know that swift foxes strongly benefit from prairie dog colonies in the northern Great Plains ( Dark Smiley and Keinath 2003, Parker et al. 2019), an d studies suggest that prairie dog colonies provide a major source of food throughout the southern portion of the species range (Kintigh and Anderson 2005). Burrowing owls may represent a competing predator to swift fox, who together , share a prey resource in mountain plover (Parker et al. 2019). To this extent; however, we cannot say with any certainty that colony size or associated species presence influences the probability of occupancy of a swift fox on TBNG, and we allude that th ese inconclusive result s reflect the sparse nature of our data set. A weak but positive effect of colony size influenced the probability that a prairie dog colony is occupied by a burrowing owl over a pooled sampling size, n=88. This finding is

PAGE 45

37 consistent with studies occurring in Phillips County, Montana, which suggest that colony size is a determining factor in owl occupancy (Alverson and Dinsmore 2014). We had n=34 presence records for burrowing owl over our three year sampling period, which is still a sparse representation w ithin the data set (Table 2a). This is a likely explanation for the poor model convergence in the remaining models within the set. We know that burrowing owls are strongly tied to prairie dog colonies where owl and prairie dog ranges overlap, and that burr ow systems brood rearing requirements (Poulin et al. 2011, Bryan and Wunder 2014). Additionally, the food resources that prairie dog coloni es offer provide an exc ellent source of forage opportunity for owls (Conrey 2010). We cannot present with any certainty; however, that owls respond to the presence of swift fox or mountain plover on TBNG. Producing results that suggest any influence at all from an effect of asso ciated species presence on burrowing owl occupancy will require a more robust data set. A stronger positive effect of colony size was identified as an influencing factor on the probability that a prairie dog colony is occupied by a m ountain plover over a p ooled sampling size, n=88. Like burrowing owls, findings in Phillips County, Montana suggest that mountain plovers respond positively to changes in colony size (Dinsmore and Smith 2010). On TBNG, larger tracts of habitat patches are likely more attractive to breeding plovers because these areas provide the only source of habitat across the landscape (Parker et al. 2019). We had n=42 presence records of mountain plover over our pooled data set, which was a sparse enough number to preve nt proper convergence o f the remaining models in the set. We cannot say with certainty that associated species presence influences the probability that a colony is occupied by a plover on TBNG, and we reiterate that a larger data set spanning multiple year s and an

PAGE 46

38 increased samp le unit size may be a better fit for occupancy based analysis targeting multi species systems of interest. We cannot determine a threshold by which the size of a colony is maximized for predicting occupancy estimates. All models tha t estimate probability of occupancy as a function of a quadratic effect on colony size either poorly converged or produced an uninterpretable amount of noise associated with the Bayesian trace plots (Table 3a c, S2 Appendix). We had an uneven distribution in size among our colon ies sampled (Table 2b, Figure 2), which combined with a limited sample size, may explain our inability to produce a reliable set of estimates. Research designs that investigate rates of occupancy typically contain extende d periods of repeated a nnual surveying across many sampling units (MacKenzie et al. 2006). When data sets contain large sample sizes over several consecutive years, models can be constructed to estimate multiple parameters that define changes between years (MacKenzie et al. 2003 ). Because patch characteristics change between years, it makes sense biologically that the probability of associated species occupying a colony would also change. We were unable to estimate a year effect on the probability of occupa ncy of prairie dog colo nies by foxes, owls, and plovers due in part to sparse data. Extending this data set for several consecutive years may increase the likelihood of identifying a year effect on probability estimates and would be informative for targeti ng long term management objectives. We note that the models that converged within our model sets only contained estimates for a single covariate effect ( i.e . colony size); therefore, our justification for our results is that our sparse data was unable to s upport an increased num ber of parameters in our poorly converging models. Parker et al. (2019) proposed that foxes, owls, and plovers are involved in a multi species interaction, where foxes control owls through intra guild predation on TBNG. This would allow

PAGE 47

39 for populations o f a shared prey resource, mountain plover, to increase (Parker et al. 2019). US Forest Service trend data suggests that co occurrence between foxes, owls, and plovers influenced individual species presence on both TBNG and Pawnee Nat ional Grassland in nort heast Colorado (Parker et al. 2019). Our initial hypotheses address this interaction by suggesting that an effect on occupancy by one species occurs from the presence of other co occurring species on TBNG. Our limited data set preven ted our ability to expl ore a species co occurrence effect; however, these hypotheses should not be overlooked in future analyses. Our analysis does not confirm or deny that inter specific interactions are a competing hypothesis when determining probability estimates for occupanc y in patchy habitats. Instead, our analysis suggests that sparse data is a likely explanation for our negative results. Robust design multi species occupancy models that address interactions between co occurring species should targe t multi year survey eff orts with an increased number of patch units sampled (Richmond et al. 2010, Robinson et al. 2014, Broms et al. 2016, Rota et al. 2016). Data sets of this nature allow for modeling the occupancy of one species conditional upon the occ upancy of one or more s pecies, rather than testing for an effect based on presence or absence as presented in this study. We also note that occupancy may not be defined by any one covariate, and that influences from human presence, disease, vegetation stru cture and diversity, an d climate necessarily must be considered in all ecological systems. We were able to identify effects of colony size on burrowing owl and mountain plover occupancy, and we can say that as colony size increases, the likelihood that a c olony is occupied by ow ls or plover also increases. However, colony size on TBNG is determined by lethal and non lethal management, disease, and climate, and these factors influence the vegetation structure and composition. Taking these factors into consid eration will be a neces sary step if identifying interactions between species of

PAGE 48

40 conservation concern in this complex system remains a target of interest. Such research will require the logistical resources necessary to invest in a long term multi species e xperiment in an ever ch anging patchy environment.

PAGE 49

41 R EFERENCES Adler PB, HilleRisLambers J. 2008. The influence of climate and species composition on the population dynamics of ten prairie forbs. Ecology . 89:3049 3060. Alverson KM, Dinsmore SJ. 2014. Factors affe cting Burrowing Owl occupancy of prairie dog colonies. The Condor. 116:242 250. Augustine DJ, Baker BW. 2013. Associations of grassland bird communities with black tailed prairie dogs in the North American Great Plains. Conservation Biology. 27:324 334. Augustine DJ, Dinsmore SJ, Wunder MB, Dreitz VJ, Knopf FL. 2008. Response of mountain plovers to plague driven dynamics of black tailed prairie dog colonies. Landscape Ecology . 23:689 697. Augustine DJ, Matchett MR, Toombs TP, Cully Jr. JF, Johnson TL, S idle JG . 2008. Spatiotemporal dynamics of black tailed prairie dog colonies affected by plague. Landscape Ecology . 23:255 267. Augustine DJ, Skagen SK. 2014. Mountain plover nest survival in relation to prairie d og and fire dynamics in shortgrass steppe. Journal of Wildlife Management . 78:595 602. Bailey L, Adams M. 2005. Occupancy models to study wildlife. USDI Geological Survey. Patuxent Wildlife Research Center; Forest and Rangeland Ecosystem Science Center. Broms KM, Hooten MB, Fitzpatrick RM. 2016 . Model selection and assessment for multi species occupancy models. Ecology. 97:1759 1770. Bryan R, Wunder MB. 2014. Western burrowing owls ( Athene cunicularia hypugaea ) eavesdrop on alarm calls of black tailed prairie dogs ( Cynomys ludovicianus ). Etho logy. 120:180 188. Butts KO, Lewis JC. 1982. The importance of prairie dog towns to burrowing owls in Oklahoma. Oklahoma Cooperative Wildlife Research Unit. Oklahoma State University, Stillwater, Oklah oma. Chi D. 2016. How a former Phoenix landfill becam e home for displaced burrowing owls. Audubon. (1 March 2018; www.audubon.org/news/how former phoenix landfill became home displaced burrowing owls ) . Colorado Parks & Wildlife (CPW). 2016. Furbearer management report: 2015 2016 harvest year. Colorado Parks & Wildlife (CPW). 2015. State wildlife action plan: A strategy for conserving wildlife in Colorado.

PAGE 50

42 Conrey R. 2010. Breeding success, prey use, and mark resight estimation of burrowing owls nesting on black tailed prairie dog towns: Plague affects a non susceptible raptor (Dissertation for Doctor of Philosophy). Digital Collections of Colorado, Colorado State University, Fort Collins. htt ps://dspace.library.colostate.edu/handle/10217/39036 . Coupland RT. 1961. A reconsideration of grassland classification in the Norther n Great Plains of North America. Journal of Ecology . 49:135 167. Crooks KR, Soulé ME. 1999. Mesopredator release and avifaunal extinctions in a fragmented system. Nature . 400:563 566. Cully JF Jr, Johnson TL, Collinge SK, Ray C. 2010. Disease limits populations: Plague and black tailed prairie dogs. Vector Borne and Zoonotic Diseases. 10:7 15. Dark Smiley DN, Keinath DA. 2003. Species assessment for swift fox ( Vulpes velox ) in Wyoming. USDA Buraeu of Land Management. Wyoming State Office. Desmond MJ, Savidge JA, Eskridge KM. 2000. Correlations between burrowing owl and black tailed prairie dog declines: A 7 year analysis. Journal of Wildlife Management . 64:1067 1075. Dinsmore SJ, Smith MD. 2010. Mountain plover responses to plague in Montana. Vector Borne and Zoonotic Diseases. 10:37 45. Dinsmore SJ, White GC, Knopf FL . 2005. Mountain plover population responses to black tailed prairie dogs in Montana. Journal of Wildlife Management . 69:1546 1553. Duchardt CJ, Porensky LM, Augustine DJ, Beck JL. 2018. Disturbance shapes avian communities on a grassland sag ebrush ecotone . Ecosphere. 9:e02483. Eads DA, Biggins DE. 2017. Paltry past precipitation: Predisposing prairie dogs to plague? Journal of Wildlife Management . 81:990 998. Eads DA, Biggins DE, Bowser J, McAllister JC, Griebel RL, Childers E, Livieri TM, Painter C, St erling Krank L, Bly K. 2018. Resistance to deltamethrin in prairie dog ( Cynomys ludovicianus ) fleas in the field and in the laboratory. Journal of Wildlife Diseases. 54:745 754. Elmhagen B, Rushton SP . 2007. Trophic control of mesopredators in terrestria l ecosystems: Top down or bottom up? Ecology Letters . 10:197 206. Estevo CA, Nagy Reis MB, Nichols JD. 2017. When habitat matters: Habitat preferences can modulate co occurrence patterns of similar sympatric species. PlosOne. 12:e0179489.

PAGE 51

43 Field A, Sedivec K, Hendrickson J, Johnson P, Geaumont B, Xu L, Gates R, Limb R. 2016. Effects of short term cattle exclusion on plant community composition: Prairie dog and ecol ogical site influences. Rangelands. 38:34 37. Fitzgerald JA, Pashley DN, P ardo B . 1999. Bird conservation plan for the northern mixed grass prairie (Physiographic Area 37). Partners in Flight (4 March 2018; https://www .partnersinflight.org/wp content/uploads/2017/02/PA 37 Northern Mixed Grass Prairie .pdf). Grant TA, Shaffer TL, Madden EM, Nenneman MP. 2017. Contrasting nest survival patterns for ducks and songbirds in Northern mix ed grass prairie. Journal of Wildlife Management . 81:641 651. Grinde AR, Niemi GJ. 2016. Influence of landscape, habit at, and species co occurrence on occupancy dynamics of Canada Warblers. The Condor. 2016. 188:513 531. Hobbs NT, Hooten MB. 2015. Baye sian models: A statistical primer for ecologists. Princeton University Press. Hoogland JL. 2006. Conservation of the bla ck western grasslands. Island Press. Hurlbert SH. 1984. Pseudoreplication and the design of ecological field experiments. Ecological Monographs. 54:187 211. Jiang L, Joshi H, Patel SN. 2009. Predation alters rel ationships between biodiversity and temporal stability. The American Naturalist . 173:389 399. Johnson TL, Cully JF J r, Collinge SK, Ray C, Frey CM, Sandercock BK. 2011. Spread of plague among black tailed prairie dogs is associated with colony spatial cha racteristics. The Journal of Wildlife Management. 75:357 368. Kamler JF, Ballard WB, Fish EB, Lemons PR, Mote K, Perchellet CC. 2003. Habitat use, home ranges, and survival of swift foxes in a fragmented landscape: Conservation implications. Journal of Ma mmalogy . 84:989 995. Karki SM, Gese EM, Klavetter ML. 2007. Effects of coyote population reduction on swift fox demographics in Southeastern Colorado. Journal of Wildlife Management . 71:2707 2718. Kintigh KM, Anderson MC. 2005. A den centered analysis of swift fox ( Vulpes velox ) habitat characteristics in Northeastern Ne w Mexico. The American Naturalist. 154:229 239. Kitchen AM, Gese EM, Schauster ER. 1999. Resource partitioning between coyotes and swift foxes: Space, time, and diet. Canadian Journal of Zoology . 77:1645 1656.

PAGE 52

44 Lauenroth WK, Burke IC. 2008. Ecology of the shortgrass steppe: A long term perspective. Oxford University Press, Inc. Lauenroth WK, Burke IC, Gutmann, M.P. 1999. The structure and function of ecosystems in the central North Ame rican grassland region. Great Plain s Research . 9:223 259. Lindeman RL. 1942. The trophic dynamic aspect of ecology. Ecology . 23:399 417. Lonsinger RC, Gese EM, Bailey LL, Waits LP . 2017. The roles of habitat and intraguild predation by coyotes on the sp atial dynamics of kit foxes. Ecosphere . 8(3):e01749.10.1002/ecs2.1749. Macías Duarte A, Conway CJ. 2015. Distributional changes in the western burrowing owl ( Athene cunicularia hypugaea ) in North America from 1967 to 2008. Journal of Raptor Research . 49:7 5 83. MacKenzie DI, Nichols JD, Hines JE, Knutson MG, Franklin AB. 2003. Estimating site occupancy, colonization, and local extinction when a species is detected imperfectly. Ecology. 84:2200 2207. MacKenzie DI, Nichols JD, Royle JA, Pollock KH, Bailey L L, Hines JE. 2006. Occupancy estimation and modeling: Inferring patterns and dynamics of species occurrence. Academic Press. Manning JA . 2011. Factors affecting detection probability of burrowing owls in southwest agroecosystem environments. Journal of Wi ldlife Management . 75:1558 15 67. Martin DJ , White GC, Pusateri FM. 2007. Occupancy rates by swift fox (Vulpes velox) in eastern Colorado. The Southwestern Naturalist . 52:541 551. Martinson HM, Fagan WF, Denno RF. 2012. Critical patch sizes for food web modules. Ecology. 93:1779 178 6. Miller BJ, Knopf FL. 1993. Growth and survival of mountain plovers. Journal of Field Ornithology. 64:500 506. Nicholson KL, Ballard WB, McGee BK, Surles J, Kamler JF, Lemons PR . 2006. Swift fox use of black tailed prairie dog towns in Northwest Texas. The Journal of Wildlife Management . 70:1659 1666. Northrup JM, Gerber BD. 2018. A comment on priors for Bayesian occupancy models. PlosOne. 13:e0192819. Outhwaite CL, Chandler RE, Powney GD, Collen B, Gregory RD, Issac NJB . 2018. Prior specification in Bayesian occupancy modeling improves analysis of species occurrence data. Ecological Interactions. 93:333 343.

PAGE 53

45 Paine RT . 1980. Food webs: Linkage, interaction strength and community infrastructure. Journal of Animal Ecology . 49:666 685. Parker RP, Duchardt CJ, Dwyer AM, Painter C, Pierce AK, Michels TJ, Wunder MB. 2019. Trophic ecology warrants multi species management in a grassland setting: Proposed species interactions on black tailed prairie dog colonies. Rangelands. *** . ***. Plumb RE, Knopf FL, Anderson SH . 2005. Minimum population size of mountain plovers breeding in Wyoming. The Wilson Bulletin . 117:15 22. Polis GA, Holt RD. 1992. Intraguild predation: The dynamics of complex trophic interactions. Trends in Ecolog y & Evolution . 7:151 154. Polsky C, Easterling III WE . 2001. Adaptation to climate variability and change in the US Great Plains: A multi scale analysis of Ricardian climate sensitivities. Agriculture, Ecosystems & Environment . 85:133 144. Poulin RG, Tod d LD, Haug EA, Millsap BA, Martell MS. 2011. Burrowing owl ( Athene cunicularia ), version 2.0. In The Birds of North America (Poole AF, Editor). Cornell Lab of Ornithology, Ithaca, NY, USA. https://doi.org/10.21 73/bna.61 . Prugh LR, Stoner CJ, Epps CW, Bean WT, Ripple WJ, Laliberte AS, Brashares JS. 2009. The rise of the mesopredator. BioScience . 59:779 791. Ramsdell PC, Sorice MG, Dwyer AM. 2016. Using financial incentives to motivate conservation of an at ris k species on private lands. Environmental Conservation . 43:34 44. Richmond OMW, Hines JE, Beissinger SR. 2010. Two species occupancy models: A new parameterization applied to co occurrence of secretive rails. Ecological Applications. 20:2036 2046. Rippl e WJ, Estes J A, Schmitz OJ, Constant V, Kaylor MJ, Lenz A, Motley JL, Self KE, Taylor DS, Wolf C. 2016. What is a trophic cascade? Trends in Ecology & Evolution . 31:842 849. Robinson QH, Bustos D, Roemer GW. 2014. The application of occupancy modeling to evaluate intr aguild predation in a model carnivore system. Ecology. 95:3112 3123. Rota CT, Ferreira MAR, Kays RW, Forrester TD, Kalies EL, McShea WJ, Parsons AW, Millspaugh JJ. 2016. A multispecies occupancy model for two or more interacting species. Meth ods in Ecolog y and Evolution. 7:1164 1173. Royle JA, Kéry M. 2007. A Bayesian state space formulation of dynamic occupancy models. Ecology. 88:1813 1823.

PAGE 54

46 Samson F, Knopf F, 1994. Prairie Conservation in North America. BioScience . 44:418 421. Sauer JR, Blank PJ, Zipkin EF, Fallon JE, Fallon FW. 2013. Using multi species occupancy models in structured decision making on managed lands. The Journal of Wildlife Management. 77:117 127. Sauer JR, Link WA, Fallon JE, Pardieck KL, Ziolkowski Jr . DJ. 2013. The North American breeding bird survey 1966 2011: Summary analysis and species accounts. North American Fauna . 79:1 32. Schmidt KA. 2001. Site fidelity in habitats with contrasting levels of nest predation and brood parasitism. Evolutionary E cology Research . 3:633 648. Shake CS, Moorman CE, Riddle JD, Burchell II MR. 2012. Influence of patch size and shape on occupancy by shrubland birds. The Condor. 114:268 278. Smith MD, Conway CJ, Ellis LA . 2005. Burrowing owl nesting productivity: A comp arison between artificial and natural burrows on and off golf courses. Wildlife Society Bulletin . 33:454 462. Soulé ME, Estes JA, Miller B, Honnold DL. 2005. Strongly interacting species: Conservation policy, management, and ethics. BioScience . 55:168 17 6. Stratman M. 2017. Status of swift fox in eastern Colorado. Colorado Parks & Wildlife. Stukel ED. 2017. Swift fox conservation team: Report for 2015 2016. Wildlife Division Report No. 2017 04, SD Department of Game, Fish and Parks, Pierre, SD, USA. Su YS, Yajima M . 2015. 7. https://CRAN.R project.org/package=R2jags . Thiele JP, Bakker KK, Dieter CD. 2013. Multiscale nest site selection by burrowing owls i n western South Dakota. The Wilson Journal of Ornithology . 125:763 774. Thompson CM, Augustine DJ, Mayers DM. 2008. Swift fox response to prescribed fire in shortgrass steppe. Western North American Naturalist . 68:251 256. Ul ev E. 2007. Cynomys ludovicia nus. In: Fire Effects Information System, [Online]. U.S. Department of Agriculture, Forest Service, Rocky Mountain Research Station, Fire Sciences Laboratory (Producer). www.fs.fed.us/ database/feis/mammal/cylu/all.html . USDA Forest Service. 2017. FSM 2600 Wildlife, fish, and sensitive plant habitat management. Chapter 2670 Threatened, endangered and sensitive plants and animals. Forest Service Manual. Rocky Mountain Region, Denver , CO.

PAGE 55

47 USDA Forest Service. 2001. Land and resource management plan for the Thunder Basin National Grassland: Medicine Bow Routt National Forest Rocky Mountain Region. Medicine Bow Routt National Forests and Thunder basin Natio nal Grassland. U.S. Fish and Wildlife Service (USFWS). 2002. Species assessment and listing priority assignment form. US Fish and Wildlife Service website: http s://www.fws.gov/southdakotafie ldoffice/BTPD_RO_Cand.%20FINAL_12Aug04.pdf U.S. Forest Service (USFS). 2016a. Pawnee National Grassland. U.S. Forest Service (USFS). 2016b. Thunder Basin National Grassland. Douglas Ranger District. Virchow DR, Hygnstrom SE . 2002. Distribution and abu ndance of black tailed prairie dogs in the great plains: A historical perspective. Great Plains Research . 12:197 218. Webb CT, Brooks PB, Gage KL, Antolin MF. 2006. Classic flea borne transmission does not drive plague epizootics in prairie dogs. Proceed ing of the National Academy of Sciences of the United States of America . 103:6236 6241. Welsh A H, Lindenmayer DB, Donnelly CF. 2013. Fitting and interpreting occupancy models. PlosOne. 8:e52015. Whicker AD, Detling JK. 1988. Ecological consequences of p rairie dog disturbances. BioScience 38:778 785. White AM, Zipkin EF, Manley PN, Schlesinger MD. 2013. Conservation of avian diversity in the Sierra Nevada beyond a single species management focus. PlosOne. 8:e63088. Williford D, Woodin MC, Skoruppa MK. 2 009. Factors influencing selection of road culverts as winter roost sites by western burrowing o wls. Western North American Naturalist . 69:149 154. Wunder MB, Knopf FL, Pague CA . 2003. The high elevation population of mountain plovers in Colorado. The Condor . 105:654 662. Wyoming Game & Fish Department (WGFD). 2017. State Wildlife Action Plan. Ye X, Skidmore AK, Wang T. 2013. Within patch habitat quality determines the resilience of specialist species in fragmented landscapes. Landscape Ecology. 28:135 147. Zuckerberg B, Ribic CA, McCauley LA. 2018. Effects of temperature and precipitation on grassland bird nesting success as mediated by patch size. Conservation Biology. 32:872 882.

PAGE 56

48 APPENDIX A. Chapter II Additional Tables and Figures Table A I. Exhaustive list of models created to estimate a year effect, along with the effect of co occurring species presence and colony size on the probability that a prairie dog colony is occupied by swift fox, burrowing owl, and mountain plover. Here, letter a represents one of the three species, while letters b a nd c represent the combined three survey presence/absence scores for the two remaining species.

PAGE 57

49 Table A II. Analyzed model sets for a. Swift Fox, b. Burrowing Owl, and c. Mountain Plover. All models contain a year covariate. Other covariates included are colony size (area), a quadratic effect on colony size (area 2 ), and the presence of swift fox (sf), burrowing owl (bo), and mountain plover (mp) and an interaction effect ( *) between two of the species to estimate a probability of occupancy ( ). Standard deviation and Lower and upper 95% credible intervals are provided for the mean probability estimates of .

PAGE 58

50 Figure A I a f. Bayesian trace plots: Swift Fox a. Model: (.) P (.)

PAGE 59

51 b. Model: (area) P (.)

PAGE 60

52 c. Model: (area + area 2 ) P (.)

PAGE 61

53 d. Model: ((mp * bo) + mp + bo) P (.)

PAGE 62

54 e. Model: ((mp * bo) + mp + bo + area) P (.)

PAGE 63

55 f. Model: ((mp * bo) + mp + bo + area + area 2 ) P (.)

PAGE 64

56 Figure A II a f. Bay esian trace plots: Burrowing Owl a. Model: (.) P (.)

PAGE 65

57 b. Model: (area) P (.)

PAGE 66

58 c. Model: (area + area 2 ) P (.)

PAGE 67

59 d. Model: ((mp * sf ) + mp + sf ) P (.)

PAGE 68

60 e. Model: ((mp * sf ) + mp + sf + a rea) P (.)

PAGE 69

61 f. Model: ((mp * sf ) + mp + sf + area + area 2 ) P (.)

PAGE 70

62 Figure A III a f. Bayesian trace plots: Mountain Plover a. Model: (.) P (.)

PAGE 71

63 b. Model: (area) P (.)

PAGE 72

64 c. Model: (area + area 2 ) P (.)

PAGE 73

65 d. Model: (( bo * sf ) + bo + sf ) P (.)

PAGE 74

66 e. Model: (( bo * sf ) + bo + sf + area) P (.)

PAGE 75

67 f. Model: (( bo * sf ) + bo + sf + area + area 2 ) P (.)