Citation
Identifying and quantifying key drivers of nest survival in shortgrass steppe birds

Material Information

Title:
Identifying and quantifying key drivers of nest survival in shortgrass steppe birds
Creator:
Carver, Amber Rose
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
Committee Members:
Tomback, Diana
Augustine, David

Notes

Abstract:
Here I present the results of two seasons of master’s-level research on breeding birds at the Central Plains Experimental Range in northern Colorado. This research was carried out through the University of Colorado Denver, in cooperation with the U.S.D.A. Agricultural Research Service and the Rocky Mountain Bird Observatory (Now: Bird Conservancy of the Rockies), under a grant from the Nebraska Game and Parks Commission. The intent of the grant was to obtain data that would inform management of habitat for birds that breed on the North American shortgrass steppe. An emphasis was placed on increasing our understanding of best-management practices for species of conservation concern. My project was composed of two units: a nest-survival analysis pertaining to the three dominant ground-nesting species at the site, and an assessment of remotely-sensed data for explaining mortality in several species affected by a severe hail-storm. My findings inform management by underscoring the importance of vegetation as a defining characteristic of breeding niches in shortgrass steppe birds. My findings suggest the importance of nest-site vegetation to nest survival in Lark Bunting (Calamospiza melanocorys). I presents preliminary data on two other species with sample sizes too small for comprehensive analysis: Horned Lark (Eremophila alpestris) and McCown’s Longspur (Rhynchophanes mccownii). The latter is a species of conservation concern in Colorado and Nebraska. My research improves our understanding of the influences to population declines in locally threatened species. Research behind my second chapter informs monitoring by establishing the strong connection between remotely-sensed weather data and weather-induced bird mortality. Thus, the second chapter of my thesis satisfies the mandate to identify tools that can be used to inform and improve management.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
Copyright Amber Rose Carver. 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
IDENTIFYING AND QUANTIFYING KEY DRIVERS OF NEST SURVIVAL IN
SHORTGRASS STEPPE BIRDS by
AMBER ROSE CARVER BS/BA, The Evergreen State College, 2008
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 Integrative Biology
2016


This thesis for the Master’s degree by Amber Rose Carver has been approved for the Department of Integrative Biology by
Michael Wunder, Advisor Diana Tomback, Advisor
David Augustine


Carver, Amber (MS, Biology)
Identifying and Quantifying Key Drivers of Nest Survival in Shortgrass Steppe Birds Thesis directed by Professor Michael Wunder and Professor Diana Tomback
ABSTRACT
Here I present the results of two seasons of master’s-level research on breeding birds at the Central Plains Experimental Range in northern Colorado. This research was carried out through the University of Colorado Denver, in cooperation with the U.S.D.A. Agricultural Research Service and the Rocky Mountain Bird Observatory (Now: Bird Conservancy of the Rockies), under a grant from the Nebraska Game and Parks Commission. The intent of the grant was to obtain data that would inform management of habitat for birds that breed on the North American shortgrass steppe. An emphasis was placed on increasing our understanding of best-management practices for species of conservation concern. My project was composed of two units: a nest-survival analysis pertaining to the three dominant ground-nesting species at the site, and an assessment of remotely-sensed data for explaining mortality in several species affected by a severe hail-storm. My findings inform management by underscoring the importance of vegetation as a defining characteristic of breeding niches in shortgrass steppe birds. My findings suggest the importance of nest-site vegetation to nest survival in Lark Bunting (Calamospiza melanocorys). I presents preliminary data on two other species with sample sizes too small for comprehensive analysis: Horned Lark (.Eremophila alpestris) and McCown’s Longspur (Rhynchophanes mccownii). The latter is a species of conservation concern in Colorado and Nebraska. My research improves our understanding of the influences to population declines in locally threatened species. Research behind my second
m


chapter informs monitoring by establishing the strong connection between remotely-sensed weather data and weather-induced bird mortality. Thus, the second chapter of my thesis satisfies the mandate to identify tools that can be used to inform and improve management.
The form and content of this abstract are approved. I recommend its publication.
Approved: Michael Wunder and Diana Tomback
IV


ACKNOWLEDGEMENTS
I thank the Colorado Field Ornithologists and the Denver Field Ornithologists for funding both years of my project. I thank Dr. Susan Skagen from the United States Geologic Survey for her material and intellectual support of the project. I thank my field assistants Syed Asif, Moriah Bell, Aaron Yappert, Arielle McDermott-Amos and David DeSimone for their hard work and dedication.
v


TABLE OF CONTENTS
CHAPTER
I RELATIVE IMPORTANCE OF VEGETATION AND WEATHER TO THE NEST SURVIVAL
OF GROUND-NESTING SHORTGRASS STEPPE PASSERINES.........................................1
Abstract.............................................................................1
Introduction.........................................................................2
Species Niche and Range Models....................................................3
Importance of Nest Survival.......................................................5
Causes of Grassland Bird Nest Failure.............................................8
Mediators of Nest Depredation.....................................................9
Nest-site Niche and Survival.....................................................11
Case Example: Birds of the Shortgrass Steppe.....................................12
Objectives.......................................................................18
Hypotheses.......................................................................18
Methods.............................................................................19
Study Area.......................................................................19
Nest Location and Monitoring.....................................................20
Nest Site Vegetation and Weather.................................................21
Basic Analyses...................................................................22
Objective A: Niche Delineation...................................................22
Objective B: Drivers of Nest Survival............................................23
Results.............................................................................23
Nest Site Vegetation and Weather.................................................25
Objective A: Niche Delineation ..................................................27
vi


Objective B: Drivers of Nest Survival
29
Discussion.....................................................................33
II WEATHER RADAR DATA CORRELATE TO HAIL-INDUCED MORTALITY IN
GRASSLAND BIRDS..................................................................35
Abstract.......................................................................35
Introduction...................................................................36
Methods........................................................................40
Study Area...................................................................40
Nest Site Measurements.......................................................41
Data Analysis................................................................43
Results........................................................................45
Discussion.....................................................................48
LITERATURE CITED...............................................................53
vii


CHAPTER I
RELATIVE IMPORTANCE OF VEGETATION AND WEATHER TO THE NEST SURVIVAL OF GROUND-NESTING SHORTGRASS STEPPE PASSERINES
Abstract
Grassland birds avoid competitive exclusion by occupying niches that differ in space use, diet, and activity budget. Here, I compare two models for grassland bird niche structure and describe the biological relevance of these models. This includes a mechanistic model and an empirical model. The mechanistic model suggests that vertical space divergence is negligible in grassland birds. By contrast, the empirical model suggests that divergence in space in both dimensions (horizontal and vertical) is important. Nest-site selection is an example of a niche parameter that has both horizontal and vertical components. The way that we study and quantify the importance of nest-site vegetation attributes such density and composition has changed, and enhanced computing power makes it possible to fit complex models to a probability-based metric of reproductive outcome. Nest failure is primarily caused by predation, but this is rarely measured directly. Instead, features such as vegetation and weather that mediate predation are more frequently measured. On the shortgrass steppe, three ground-nesting passerines overlap in habitat preference and may dominate the local breeding bird community: Lark Bunting (LARB), Horned Lark (HOLA), and McCown’s Longspur (MCLO). All three are open-cup ground-nesting passerines, but the birds differ in several ways. Based on a literature review, I conclude that both LARB and MCLO are nomadic, while HOLA is more site-faithful. Previous research suggests that HOLA has a broader niche than LARB or MCLO. In 2014,1 located 207 nests where vegetation was later measured: 167 LARB, 21 HOLA, and 19 MCLO. I found that cover by tall vegetation (midgrass and forbs)
1


was significantly higher for LARB nests than for the other two species, and only LARB nests were associated with shrubs. Contrary to expectation, LARB appeared to occupy a broader niche than the other two species, with regard to nest-site vegetation. I used 253 nests that did not fail due to the direct impact of weather (200 LARB, 28 HOLA, and 25 MCLO) to fit time- and weather- based models, looking at both immediate and longer-term impacts of precipitation and temperature. I concluded that survival for all species increased with successive nest age, decreased with time, and was negatively but inconsistently influenced by daily precipitation. For a subset of 152 nests where vegetation had been measured (119 LARB, 17 HOLA, and 19 MCLO), I fit models based on nest-site vegetation height and composition. In LARB, crown cover by dead vegetation and shrubs and cacti, as well as basal cover by bare ground, had similar moderate support from the data, and support for daily precipitation was comparatively negligible. In HOLA and MCLO, there was no clear support for any vegetation variable, which may be a reflection of small sample sizes. Additional research on these species will improve our understanding of the link between niche and the cost associated with phenotypic variability.
Introduction
Grassland bird communities have relatively low species richness but high representation by a few grassland-dependent species (Rotenberry and Wiens 1980; Knopf 1996). Understanding the features that underpin the composition of these communities is challenging, because these communities tend to be highly dynamic as a result of environmental stochasticity driven by drought, wildfire, and grazing. Species niche and range delineation have been modeled in various ways (Cody 1968; Rotenberry and Wiens 1980; Knopf 1996). Here, I compare two of these models and highlight birthrate as a demographic parameter of interest. I offer an
2


alternative model explaining the relationship between niche breadth, parent bird phenotype, and nest survival, and I assess the value of this model using two seasons of data on the nest survival of shortgrass steppe passerines.
Species Niche and Range Models
Both mechanistic and empirical models have been used to explain niche delineation and species distribution in grassland birds. Cody (1968) proposed a mechanistic model suggesting that birds avoid competitive exclusion through horizontal and vertical divergence in habitat use, diet specialization, and differential activity budget (Fig. 1). He contended that divergence in the use of vertical habitat is negligible in grassland birds due the near lack of vertical structure, allowing us to focus on the other three dimensions: horizontal space, diet, and activity. This model is useful in that it allows us to categorize likely modes of niche
Fig. 1. Cody’s (1968) mechanistic model for how bird species avoid competitive exclusion, with arrows representing direction of selective
3


divergence, but it is based more on deductive reasoning than on empirical observation and overlooks many of the complexities associated with dynamic grassland bird communities.
In contrast, Knopf (1996) presented an empirical model for the distribution of birds on the Great Plains (Fig. 2). This model is informative in that it synthesizes many observations. Unlike the earlier model, it does not make the assumption that vertical structure is unimportant in grassland systems. Rather, it recognizes the existence of ephemeral emergent vegetation whose distribution is driven in part by spatiotemporally variable grazing pressure. Furthermore, it includes niche breadth and overlap, providing a more complex and biologically precise representation of how Great Plains species are distributed. It provides a framework for considering the requirements of bird species coexisting in a heterogeneous landscape. While it is limited to birds treated by the model, it could easily be extended based
Excessive
Moderate

Light -^+Jono

â–  Mountain Plover---
- McCown's Langspur ----Ferruginous Hawk
"Uirtg-kjiliod Curlew---------1
> Lark Burning â– 
— Chestnut-collared Longspun------i
I----------Sprague's Pipri--------
â–  Bard'a Sparrow -
Cassm's Sparrow
Bare --------------- Short
Mixed---------—Mixedffihrub
Fig. 2. Knopf s (1996) model for the distribution of Great Plains birds. This empirical model demonstrates the concept that species distribution responds to vegetation structure, which in him is a factor of grazing intensity. Thus, arazina-induced habitat heteroaeneitv reaulates communitv coniDosition in Great Plains birds.
4


on our understanding of other species’ life histories. Its primary limitations are that it focuses exclusively on the grazing-vegetation relationship and does not show how communities respond to change over the course of time.
One of the key features not addressed by either model is the cost of individual phenotype. Population genetic variability results in phenotypic heterogeneity, and individual reproductive fitness in a given set of environmental conditions depends in large part on phenotype. This selective pressure means that individuals should generally be found in the environment where they will be reproductively successful (Adams 1908; Grinnell 1917). If optimal habitat is limited, some individuals will occupy suboptimal habitat, although survival declines with departure from optimal conditions (Fig. 3, Holt and Keitt 2005). Niche breadth is regulated in part by the reproductive penalty associated with deviance from mean phenotype (Slatyer et al. 2013). In the case of nest-site selection, the probability of producing at least one offspring from a nesting effort (nest survival) should decrease with departure from mean nesting conditions, which would include features such as vegetative cover, proximity to other nesting pairs, effort timing, and food availability. Thus, quantifying the mean and variance of nesting conditions for each species can inform us as to the biology of each species and the modes of niche divergence. Modeling nest survival dynamics is important not only to understanding how populations respond to habitat, but also to understanding the link between population demography and community flux.
Importance of Nest Survival
Population demography and stability are regulated by birth, death, immigration, and emigration (Pulliam 1988). Of these four demographic parameters, birth rate is the most
5


tractable problem in grassland birds for two reasons. First, breeding season adult distribution is predictable based on species niche (Johnson and Grier 1988). Although individuals may have low site fidelity (Jones et al. 2007), species occupancy of a given area can be predicted based on habitat attributes (e.g., Pons et al. 2003; Irvin et al. 2013; Mclaughlin et al. 2014). Second, nests can be located based on adult distribution. Once located, nests can be
monitored until effort completion, and fate can The most challenging aspect of this form of research is deriving a biologically relevant response metric for nest-site habitat attributes.
Until the early 1960’s, the prevailing metric of avian reproductive outcome was apparent nest success (e.g., Peterson and Young 1950; Glover 1953; Glover 1956; Labisky 1957; Nice 1957; Steel et al. 1957). Apparent nest success is defined as the number of known nesting efforts in which at least one offspring survived and left the nest. Mayfield (1961,
1975) was the first to publicly point out that apparent nest success is a poor metric of population reproductive response. This is because most methods of nest location rely on observing adult behavior, and adults do not remain near the nest site after a nest has failed.
ascertained with a fair degree of certainty.
Fig. 3. Conceptual model relating species distribution to demographic parameters (Holt and Keitt 2005). If
(a) density-independent mortality increases across a gradient in habitat quality (xi to X3), density-dependent birthrate should increase, suggesting that
(b) habitat quality (and hence carrying capacity) is lowest in xi.
6


Many nests fail early in the nesting effort, so most nest samples are skewed toward nests that avoid failure longer. Mayfield advocated for the use of a probability-based metric—now termed nest survival (Dinsmore et al. 2002)—that quantifies the likelihood of achieving success rather than the proportion of nests known to have succeeded.
Although Mayfield (1961) was unquestionably influential (Google Scholar citations= 1,596), researchers were slow to adopt his methods and terminology during the first 40 years following publication (Fig.4). Since the late 1990’s, the annual numbers of publications including the term “nest survival” (alone or in combination with “nest success”) and “nest success” (alone) have both grown dramatically. This indicates an increased research interest likely fueled by enhanced computing power, which allowed us to move beyond Mayfield’s overly-simplistic constant survival estimates to consider more subtle patterns in survival (Dinsmore et al. 2002; Jehle et al. 2004). Specifically, the ability to fit multivariate generalized linear models to survival data has been facilitated by automated computation through programs such as MARK (White and Burnham 1999) and scripting languages such as SAS (Rotella et al. 2004) and R (Laake and Rexstad 2014).
That so many studies still use the term “nest success” without referring to “nest survival” (Fig. 4) suggests either the use of outdated terminology or continued use of a metric that has been shown to have little biological relevance. The results of such studies must be interpreted with caution, as they may be poorly reflexive of actual nest survival dynamics.
7


Fig. 4. Trend in publication of nesting outcome literature. Literature including the term “nest survival” (solid line) and literature only referring to “nest success” (dashed line) have both increased with the availability of automated computation. Publication for many biological topics follows a similar pattern, but few topics have shown such a dramatic change as nest survival literature, which went from <10 publications per year for the first 30 years after Mayfield (1961), to >100 per year over the most recent decade.
Causes of Grassland Bird Nest Failure
I conducted a literature search using the term “grassland bird” AND “nest failure” and identified a sample of 173 journal articles directly pertaining to causes of grassland bird nest failure. This sample excluded reviews, meta-analyses, and research on marsh or coastal birds. I found that nearly all studies in the sample identified depredation as the primary cause of nest failure. In agricultural grasslands, other potential causes of low nest survival include trampling by livestock, destruction through mowing, and exposure to pesticides. Trampling was found to be important in some studies (Perlut et al. 2011; Beja et al. 2014; Chaiyarat and Eiam-Ampai 2014) but not others (Winter et al. 2004; Johnson et al. 2012; Bleho et al. 2014).
8


Mowing causes low nest survival in some systems (Kerschner and Bollinger 1996; Casas and Vinuela 2010) but not others (Giocomo et al. 2008). Pesticide effects are understudied.
Martin et al. (1998) found that Chestnut-collared Longspur was capable of switching diet in response to reduced abundance of a prey item after pesticide application and that nest survival was not strongly impacted. However, the prevalence of pesticide use in grasslands and concurrent idiopathic declines in grassland birds suggest a possible relationship that should be investigated further (Mineau and Whiteside 2013).
Weather is another understudied influence on nest survival. Perlut and Strong (2011) found that a high proportion of Savannah Sparrow nests failed due to flooding during heavy rain. Hartman and Oring (2009) found that prior winter precipitation explained flood-induced nest failure in Long-billed Curlew. Severe weather may directly cause low nest survival and adult mortality in birds (Wauer and Wunderle 1992; Wiley and Wunderle 1993; Newton 2007; Frederiksen et al. 2008; McKechnie and Wolf 2010; Saunders et al. 2011; McKechnie et al. 2012). However, quantifying the link between weather and nest survival is challenging because studies cannot be planned around naturally occurring severe weather events. That is because such events are unpredictable. For further discussion on this and a case example, see Chapter 2.
Mediators of Nest Depredation
Although depredation is the primary cause of nest failure in grassland birds, predator abundance and activity levels are rarely measured directly. Exceptions include Klug et al. (2010), Stanley (2010), and Murray (2015). For a review on nest camera studies, see Degregorio et al. (2014). Grassland nest predators may include mammals (Stanley 2010;
9


Murray 2015), snakes (Klug et al. 2010), and birds (MacDonald and Bolton 2008), depending on the location and habitat. Researchers more often choose to measure habitat features that mediate predation, such as nest-site vegetation (Winter et al. 2004; Warren and Anderson 2005; Stauffer et al. 2011; Johnson et al. 2012), nesting area patch size (Skagen et al. 2005; Winter et al. 2006; Weidman and Litvaitis 2011; Benson et al. 2013), patch perimeter-area ratio (Helzer and Jelinski 1999), nest proximity to habitat edge (Bollinger and Gavin 2004; Benson et al. 2013; Key el et al. 2013; Perkins et al. 2013) or to woody vegetation (Vos and Ribic 2013), nest area habitat (Henningsen and Best 2005; Lloyd and Martin 2005; Winter et al. 2006; Cao et al. 2009; Blank et al. 2011; Hatchett et al. 2013), or local management regime (Sheldon et al. 2007; Churchwell et al. 2008; With et al. 2008;
Prevalence of x (%)
Fig. 5. Hypothetical niche distribution. Assuming a range of 0-100% in prevalence of a given feature (x), nest-site placement may be independent of that feature (gray), more likely to be placed where that feature has low prevalence (blue) or high prevalence (purple). Nests of species with higher niche breadth (generalists) with respect to feature x should be more variable in regards to prevalence of x (orange). For variables important to survival, mean prevalence of x should correspond with highest nest survival and highest number of nests.
10


Johnson et al. 2012; VanBeek et al. 2014). In addition, they may measure daily weather attributes such as rainfall and temperature (Dinsmore et al. 2002; Skagen and Yackel Adams 2012; Ludlow et al. 2014), or seasonal attributes such as rainfall (Skagen and Yackel Adams
2012) or drought index (Dreitz et al. 2012), which are thought to influence predator distribution and activity level. Nesting density may also be important to survival, perhaps due to increased detection by predators in areas with higher nest density (Aguilar et al. 2008; Skagen and Yackel Adams 2010) or enhanced vigilance in a group-nesting situation (MacDonald and Bolton 2008). Relationships between environmental variables and nest survival depend on species life history (Davis et al. 2006; Kerns et al. 2010; Grant and Shaffer 2012; Lusk and Koper 2013), so nest survival analyses are an important tool for quantifying niche attributes.
Nest-site Niche and Survival
As illustrated by Cody (1968), grassland bird species niche is comprised of time budget, diet, and space use. The latter includes nest-site vegetation. Knopf (1996) highlighted one of these—nest-site vegetation—and here I focus on this aspect as well. Nest-site vegetation may strongly influence nest survival but has variable importance depending on species life history (Davis 2005; Davis et al. 2006; Kerns et al. 2010; Grant and Shaffer 2012; Lusk and Koper
2013) . Thus, elucidating the relative importance of different nest vegetation attributes provides information on niche position and breadth (Fig. 5). The range of nest conditions used by each species—as well as the magnitude of decline in nest survival with departure from mean conditions—are key attributes of niche (Adams 1908; Grinnell 1917). Grassland ecosystems are highly dynamic, making the exact distribution of conditions unpredictable
11


from year to year (Winter et al. 2005). Most grassland breeding birds are migratory, and upon return to the breeding range they may follow any of the three settling patterns proposed by Johnson and Grier (1988): homing, opportunistic settling, or flexible settling. There is evidence to suggest that many species exhibit either opportunistic or flexible settling, resulting in perceived nomadism. In this way, they track the distribution of conditions rather than returning to the same location and suffering the population-level reproductive consequences of breeding in suboptimal conditions (Fretwell 1986; Igl and Johnson 1999; Winter et al. 2005). This lends support to the hypothesis proposed in Fig. 5, because nomadism allows species to achieve niche stability despite environmental stochasticity.
Case Example: Birds of the Shortgrass Steppe
The shortgrass steppe is a region along the western edge of the North American Great Plains where low precipitation and high evaporative potential (Sala et al. 1992) result in dominance by low-stature grasses (Milchunas et al. 1989). The ecosystem is defined by disturbance-induced habitat heterogeneity that was historically driven by a combination of grazing by bison and prairie dogs, as well as periodic drought (Demer et al. 2009). These stochastic factors occur with different periodicities and distributions, generating a complex shifting habitat pattern that is a key regulator of bird community composition (Rotenberry and Wiens 1980; Knopf 1996; Derner et al. 2009). Birds of the shortgrass steppe occupy niches that show varying overlap (Rotenberry and Wiens 1980; Knopf 1996). Here, I focus on three species: Lark Bunting (Calamospiza melanocorys), Horned Lark (Eremophila alpestris), and McCown’s Longspur (Rhynchophanes mccownii). All three belong to the order Passeriformes and are short-distance migrants within the Great Plains. Ecologically, they are
12


similar in that they build open-cup ground nests, but they differ in several important ways (Table 1). I selected these species because they breed in similar habitat, and because where they overlap in breeding range they may be the dominant ground-nesting species.
Lark Bunting is in the family Emberizidae and is thought to be highly nomadic in the breeding season. Their nomadism has not been documented directly through empirical research, but it is supported circumstantially by the widely fluctuating local abundance of Lark Bunting (Augustine and Derner 2015). This fluctuation is almost certainly due to local movements, because magnitude and frequency of change are greater than might be expected through varying rate of birth and death. Lark Bunting nomadism is also indicated by the fact that they are highly sexually dichromatic (Chaine and Lyon 2008). Increased ornamentation is a sign of stable access to habitat resources (Polak and Starmer 2005), which in the dynamic shortgrass steppe ecosystem strongly suggests that Lark Buntings track resource availability across the landscape, exhibiting opportunistic or flexible settling rather than homing. Male ornamentation is reinforced by polygyny, because females preferentially mate with males on the best territories even when those males have already mated with other females (Pleszczynska 1978). Lark Bunting occupancy is not tied to vegetation attributes (Rotenberry and Wiens 1980), suggesting that these birds track other environmental features (such as precipitation or food availability) rather than vegetation structure.
Although Lark Bunting is classified as a new world sparrow (Carson and Spicer 2003), it is different from many other shortgrass steppe sparrows in that it nests on the ground rather than in a shrub (Pleszczynska 1978). It nests in taller, denser vegetation than Horned Lark or McCown’s Longspur, and the nest is relatively well-shaded throughout the day (With and Webb 1993; Lusk and Koper 2013).
13


Lark Bunting nest survival increases with median grass height (Skagen and Yackel Adams 2010), but weather appears to exert a stronger and more consistent effect on Lark Bunting nest survival than nest-site vegetation does (Skagen and Yackel Adams 2012). Similarly, Lark Bunting post-fledging survival is negatively impacted by drought (Yackel Adams et al. 2006). In summary, Lark Bunting is a likely nomadic, polygynous species that
tends to nest in tall vegetation but whose reproductive success appears tied more to weather
Table 1. Comparison of Focal Species
Species Lark Bunting Horned Lark McCown's Longspur
Family Emberizidae Alaudidae Calcariidae
Breeding Range throughout the Great Plains throughout the northern hemisphere northwestern Great Plains
Breeding Habitat mixed grass, shortgrass, shrubland shortgrass shortgrass
Breeding Range Settling Opportunistic or flexible (probable) Homing (probable) Opportunistic or flexible (probable)
Peak Breeding on Shortgrass Steppe Late May and Mid-June Late April, early May Mid-May and Mid-June
Sex Physical Differences Males ornamented, females not Males larger than females Males ornamented, females not
Mating System polygyny Social and genetic monogamy (probable) Social monogamy, with potential genetic polyandry (probable)
Parent Roles Both build, incubate, and provision Females tend nests more than males Only females incubate, but both provision young
Nest Vegetation In midgrass, forbs, under shrub Sparse, low vegetation Sparse, low vegetation
Drivers of Nest Success Seasonal precipitation (+), daily precipitation (-), daily temperature (+), grass height (+) unknown Proximity to shrubs (-)
and climate than to nest-site vegetation.
14


Horned Lark is a Holarctic species in the family Alaudidae. The species breeds in a wide range of habitats, including polar deserts and arctic tundra (Ryzhanovsky 2015), alpine tundra (Camfield and Martin 2009), marine islands (Mason et al. 2014), desert (Dean and Williams 2004), and cropland (Beason and Franks 1974). This suggests high tolerance and adaptability, but inter-population variance in genes (e.g., Mason et al. 2014) and phenotypes (e.g., Camfield and Martin 2009) has not been comprehensively quantified. Homed Lark is categorized as year-round resident on the shortgrass steppe, but may migrate short distances which means that we may not be seeing the same individuals throughout the year (With and Webb 1993; Manegold and Sudhaus 2004). On the Great Plains, they are strongly associated with low-stature vegetation (Rotenberry and Wiens 1980). Horned Larks are physiologically adapted to temperature extremes and low water availability, which may serve as an alternate strategy to nomadism (Davies 1982; Swain 1991). Sex plumages are not distinct, suggesting a less polygynous social structure (Polak and Starmer 2005). They are, however, physically dimorphic: males are larger than females and invest less energy in nestling provisioning (Du et al. 2015). The extent of social or genetic monogamy in Horned Lark is unknown, but empirical anecdotes indicate complex social behavior (Camfield et al. 2007).
Horned Lark nests tend to be placed in extremely short vegetation with little or no nest cover, but they are frequently placed in the lee of an emergent feature such as a cactus pad, rock, or pile of dung (Dubois 1935; Beason and Franks 1974). This feature offers slight protection from wind but almost no shade, which may indicate a partial reliance on solar radiation for incubation (With and Webb 1993). Habitat patch size does not affect Homed Lark occupancy (Davis 2004) and has a slightly negative but inconsistent effect on nest survival (Skagen et al. 2005). As in Lark Bunting and McCown’s Longspur, distance from
15


patch edge does not affect nest survival in this species (Stanley 2010). Few studies have looked at nest survival dynamics in Horned Lark, so potential relationships can only be inferred from life history. Nests of this species are highly exposed and have not been documented in dense, tall vegetation. However, Rotenberry and Wiens (1980) estimated higher niche breadth with respect to vegetation for McCown’s Longspur than for Lark Bunting or McCown’s Longspur. With respect to climate, this species also occupies a broad niche. In summary, Horned Lark is a likely non-nomadic species that nests in short vegetation and whose reproductive success should be tied more to vegetation than weather.
McCown’s Longspur is a shortgrass specialist in the family Calcariidae (Klicka et al. 2003). Like Homed Lark, occurrence of this species is typically associated with shorter vegetation than would be expected based on climate alone (Rotenberry and Wiens 1980; Henderson and Davis 2014), signifying a need for grazing disturbance. It should be noted that Bogard and Davis (2014) found that McCown’s Longspur occupancy decreased with percent cover by grass and increased with vegetation height. However, the effects were modest (P=-0.022 and 0.033 respectively), and do not necessarily contradict the general consensus that McCown’s Longspur is shortgrass obligate. Occupancy is more strongly tied to grazing by large, generalist herbivores (Mickey 1943; Felske 1971) and prairie dogs (Augustine and Baker 2013) than to fire (Augustine and Derner 2015). Like Lark Bunting, the species is sexually dichromatic, with males being more strongly ornamented than females (DuBois 1937), suggesting genetic polyandry though they appear to be socially monogamous (Mickey 1943, Felske 1971). Also like Lark Bunting, McCown’s Longspur local abundance fluctuates widely between years (Martin and Forsyth 2003), indicating that this species may be nomadic as well.
16


McCown’s Longspur may breed on cropland (Martin and Forsyth 2003), but it is much more common on rangeland (Mickey 1943, Felske 1971). Breeding territories have greater coverage by low-stature vegetation, more cattle dung, more lichen, and more forbs than generally available habitat, but vegetation does not strongly affect apparent nest success (Greer and Anderson 1989). It is a semi-colonial breeder, nesting in denser aggregations than would be predicted through habitat quality alone (Mickey 1943, Felske 1971). Females are solely responsible for incubation, but both males and females provision young (DuBois 1937, Mickey 1943). Similar to Horned Lark nests, nests of McCown’s Longspur are placed in low vegetation and typically have little cover, though they are often placed in the lee of an emergent feature that provides no shade (Dubois 1935; DuBois 1937; Mickey 1943; With and Webb 1993). Apparent nest success is lower near shrubs, probably due to a higher density of egg-eating mammals in areas with perennial woody vegetation (With 1994). In summary, McCown’s Longspur is a likely nomadic, shortgrass obligate species whose nest success is influenced by woody vegetation.
Based on the above species profiles, I hypothesize that both Lark Bunting and McCown’s Longspur are nomads. Lark Bunting tracks changes in the distribution of food and/or weather, while McCown’s Longspur tracks changes in the distribution of grazing pressure and the associated vegetation. I hypothesize that Horned Lark is not nomadic and instead relies on physiological adaptations to cope with fluctuations in local breeding conditions. Homed Lark occupies a much broader niche than Lark Bunting or McCown’s Longspur (Rotenberry and Wiens 1980), an adaptation that compensates for being non-nomadic.
17


Objectives
A) Quantify differences in nest-site vegetation height and composition in the three dominant ground-nesting species at the site: Lark Bunting, Horned Lark, and McCown’s Longspur.
B) Quantify relative importance of recent weather—including temperature and precipitation—to nest survival in each species.
C) Quantify relative impact of vegetation characteristics on the nest survival of each species.
Hypotheses
A) Lark Bunting nests will have significantly higher percent cover by midgrasses, forbs, shrubs, and sub-shrubs than Horned Lark or McCown’s Longspur. Horned Lark nests sites will vary more in these features than Lark Bunting or McCown’s Longspur.
B) Increasing daily precipitation will be decrease nest survival for all species. Rationale due to higher humidity and an increase in odor transmission and predator activity after rainfall events (Dinsmore et al. 2002). Increasing temperature will increase nest survival for all species, because it will lower the degree of activity needed to maintain metabolic stasis and warm the nest (Kendeigh 1939). This effect will be smaller in Horned Lark, which is more physiologically adapted to temperature variation.
C) Percent cover by tall vegetation (midgrasses, forbs, shrubs, and subshrubs) will be important to nest survival for all species (Fig. 6). Rationale: increase in cover by tall
18


vegetation will lead to a decline in nest survival for Homed lark and McCown’s Longspur, but an increase in nest survival for Lark Bunting.
Methods
Study Area
Figure 6. Hypothetical distribution of nests with respect to vegetation, conceptually based on figure by Knopf (1996). Assuming a range of 0-100% in extent of a given feature, McCown’s Longspur (orange) and Homed Lark (purple) nests will be more similar to one another than they are to Lark Bunting (blue) nests. Homed Lark nests will be more variable than nests of the other two species.
This study was conducted at Central Plains Experimental Range (CPER), in Weld County, Colorado on the western edge of the Pawnee National Grassland. This is long-term grazing research facility currently administered by the U.S.D.A. Agricultural Research Service (ARS). The study area consisted of 129.5-ha pastures segregated by barbed-wire fences and
19


grazed under uniform and adaptive grazing regimes with domestic cattle. The climate is semi-arid, with most precipitation occurring in the summer months (Sala et al. 1992). Longterm climate oscillations contribute to inter-annual variability in temperature and rainfall (Rosenberg 1987). Area habitat is classified as shortgrass steppe. The area was grazed intensively by American bison {Bison bison) prior to their extirpation (Sanderson et al. 2008). They grazed in dense, mobile herds, generating a shifting mosaic of heterogeneous habitat. Thus, bird species that breed on the shortgrass steppe evolved in the context of shifting disturbance and high habitat variability (Knopf 1996; Derner et al. 2009). Prairie dogs continue to act as important grazers in the region surrounding CPER, but they have been almost entirely removed from CPER in order to isolate the effect of the livestock grazing treatment. Current vegetation at CPER is dominated by C4 shortgrasses (primarily Bouteloua gracilis and B. dactyloides), with smaller proportions of cacti, midgrasses, and forbs (Lauenroth and Sala 1992). The woody plant community includes both tail-stature shrubs and low-stature subshrubs (Lee and Lauenroth 1994).
Nest Location and Monitoring
I gathered data in 2014.1 located nests through rope-dragging (Labisky 1957), focusing my search effort on 3.3-ha permanent plots that are also used for long-term vegetation and breeding bird surveys. There are four plots in each pasture, and I included eight pastures in my search effort, for a total of 32 plots (105.6 ha). I dragged each plot once per week between May and July, for a total of 7-8 repeat visits per pasture. I also found nests opportunistically. Upon locating a nest, I marked the nest location with a wooden stake, placed ~3 m to the north to avoid disturbing the nest site. I recorded nest GPS coordinates
20


and recorded nest species and number of eggs and/or young. I estimated egg development using the egg-float method (Westerskov 1950). I estimated chick development based on extent of pins and feathers, as well as behavior (Jongsomjit et al. 2007). I returned to each nest every 2-4 days to check the status, ceasing once a nesting effort had evidently ended. If a nest became empty prior the earliest possible fledging date, if the nest lining was pulled up, or if there were egg shell fragments the nest was classified as having failed due to predation. On June 22nd, a severe hail storm caused the failure of many nests, and these were not included in the analysis for this chapter. For analyses pertaining to the storm sample, see Chapter 2.
Nest Site Vegetation and Weather
An ARS field crew measured nest-site vegetation as soon as possible after a nesting effort had ended. Nests where too much time had elapsed and the vegetation had changed were omitted from the sample. Vegetation quantification included taking Visual Obstruction Readings in four cardinal directions (Robel et al. 1970) and measuring distance to the nearest shrub if there was at least one shrub within 30 meters of the nest. We also measured percent composition by major cover type within one meter of the nest using a modified version of the point intercept method (Canfield 1941). Briefly, this involved placing a metal quadrat over the nest with the nest at the center, and with a laser beam perpendicular to the ground counting the number of intercepts by cover type along a grid within the frame. In this way, we quantified both crown and basal cover. Cover types included shortgrass, midgrass, forbs, annual grass, shrubs (height >15 cm), subshrubs (height <15 cm), dead vegetation, cactus, dung, and bare ground. We calculated percent cover by each vegetation type by dividing the
21


number of intercepts of that type by the total number of counts. We collected data on daily maximum and minimum temperature, as well as total precipitation, from an on-site weather station.
Basic Analyses
I assessed whether vegetation features were normally distributed for each species using a Shapiro-Wilks test, treating features with p > 0.05 as being normally distributed. In order to address the potential for collinearity between vegetation and time caused by plant phenology, I tested for correlation between each vegetation feature and time. In order to address the potential for inter-taxa relationships such as competition or dependence, I tested for correlation between pairs of vegetation features. For any highly correlated matched pair, I eliminated all the basal cover variable from consideration. In order to quantify immediate and delayed or cumulative weather effects, I determined minimum, maximum, and range in temperature, as well as precipitation, for the preceding day. I also calculated average minimum and maximum temperature over the three days preceding nest effort conclusion. To identify and eliminate collinearity, I looked for correlation between pairs of weather variables.
Objective A: Niche Delineation
I used a Kruskal-Wallis one-way analysis of variance by ranks to assess similarity of vegetation features that were not normally distributed for all species, while I used a one-way ANOVA to compare normally distributed features, treating features with p < 0.05 as having at least one species with a significantly different distribution. I used a post-hoc Kruskal-
22


Nemenyi Test to identify pairwise differences between species for those features deemed significantly different for at least one species through the Kruskal-Wallis or ANOVA tests, and I examined boxplots to identify direction of difference.
Objective B: Drivers of Nest Survival
For nest survival analyses, I omitted the 70 nests that had failed due to hail, which left a sample of 253 nests for time- and weather-based analyses, including 200 Lark Bunting nests, 28 Homed Lark Nests, and 25 McCown’s Longspur Nests. I first quantified the effect of weather on nest survival for each species. To do so, I initially fit a set of time-based generalized linear models using R (version 3.2.2) package RMark (Laake and Rexstad 2014), which uses the R language interface to ran program MARK (White and Burnham 1999). I assessed relative model fit with AAICc (Akaike 1974) data support for the model with weight (w), and relative support for the model with the ratio between weights of two models.
The time attributes that I included in the first set of models were a time-varying factor, a constant effect of time, and the age of the nest. I tested the fit of all possible combinations of these variables, as well as an intercept-only model. The purpose of this step was to account for variability introduced by unmeasured background variables. For all species, I used the best time-based model as a template for constructing more complex weather-based models in which I added a single additional variable, using the set of precipitation variables identified as independent. For vegetation-based analyses, my sample was further reduced to include only those where vegetation had been measured,
Results
Over the course of nine weeks (May 19-July 20), I located 319 nests belonging to five species: Lark Bunting, Horned Lark, McCown’s Longspur, Brewer’s Sparrow, and Common
23


Nighthawk. Only one nest each of the latter two species was found, so these species were omitted from analyses, and I focused on the remaining three (Table 2).
Table 2. Nest sample for the three most common species.
Lark Bunting Horned Lark McCown’s Longspur Total
Nests Located 257 33 27 319
Veg. Measured 167 21 19 207
Data norma ly distributed?
CC* shortgrass No (p<0.001) No (p=0.02) Yes (p=0.36)
CC dead veg. No (p<0.001) Yes (p=0.22) Yes (p=0.12)
BC** shortgrass No (p<0.001) Yes (p=0.12) Yes (p=0.21)
BC dead veg. Yes (p=0.19) Yes (0.81) Yes (p=0.16)
BC bare No (p<0.001) Yes (p=0.27) Yes (p=0.20)
Mean high VOR*** No (p<0.001) Yes (p=0.53) Yes (p=0.29)
Mean low VOR No (p<0.001) Yes (p=0.27) No (p=0.003)
Distance to Nearest Shrub No (p<0.001) Yes (p=0.09) Yes (p=0.05)
All other veg. characteristics No (p<0.001) No (p<0.001) No (p<0.001)
At least one species different from others?
CC Midgrass Yes (p<0.001)
CC Forbs Yes (p=0.05)
BC Midgrass Yes (p<0.001)
BC Annual Grass Yes (p=0.002)
BC Dung Yes (p=0.003)
Mean high VOR Yes (p=0.001)
Mean low VOR Yes (p=0.001)
Distance to Nearest Shrub Yes (p=0.002)
All other veg. characteristics No (p<0.001)
Pairwise similarity between species
HOLA:MCLO HOLA:LARB MCLO:LARB
CC Midgrass p=0.74 p=0.01 p<0.01
CC Forbs p=0.85 p=0.31 p=0.09
BC Midgrass p=0.85 p=0.05 p<0.01
BC Annual Grass p=0.88 p=0.09 p<0.02)
Mean high VOR p=0.93 p=0.002 p<0.001
Mean low VOR p=0.15 p<0.001 p=0.44
* Crown Cover ** Basal Cover
24


*** Visual Obstruction Reading
Nest Site Vegetation and Weather
Vegetation was measured at 65% of nests (Table 2). Crown cover abundance of shortgrass was normally distributed for McCown’s Longspur but not Horned Lark or Lark Bunting. Crown cover abundance of dead vegetation was normally distributed for Horned Lark and McCown’s Longspur, but not Lark Bunting. Basal cover abundance of shortgrass was normally distributed for Homed Lark and McCown’s Longspur, but not Lark Bunting. Basal cover of dead vegetation was normally distributed for Lark Bunting, Horned Lark, and McCown’s Longspur. Basal cover of bare ground was normally distributed for Horned Lark and McCown’s Longspur, but not Lark Bunting. Mean high VOR was normally distributed for Homed Lark and McCown’s Longspur, but not Lark Bunting. Mean low VOR was normally distributed for Homed Lark but not McCown’s Longspur or Lark Bunting. Distance to the nearest nest was normally distributed for Homed Lark and McCown’s Longspur but not Lark Bunting. No other vegetation features were normally distributed. Horned Lark and McCown’s Longspur nests had no subshrubs or shrubs within one meter of the nest. I found no correlation between time and percent cover for any vegetation cover type. With the exception of dead vegetation, crown cover features were correlated (r>0.50) with basal cover of the same type, but not with any other feature. Crown and basal cover of dead vegetation were not strongly correlated (r=0.11). Minimum and maximum daily temperature showed a weak increase over the course of the season (r=0.59 and 0.63 respectively), maintaining high day-to-day variability (Fig. 7). I found that minimum daily temperature was moderately correlated with minimum (r=0.58) and maximum (r=0.63) temperature from the prior day. It was also moderately correlated with average minimum (r=0.59) and maximum (0.60)
25


temperature over the past three days. I kept minimum daily temperature for analysis and removed the moderately correlated variables from consideration. I found that maximum daily temperature was strongly correlated (r=0.75) with daily temperature range, so I removed daily range from consideration. Daily precipitation was not correlated with any other variable. Temperature range from the prior day was moderately correlated with average temperature range over the preceding three days (r=0.63), so I removed the latter from consideration. Rainfall from the prior day was moderately correlated with average precipitation over the preceding three days (r=0.61), so I removed the latter from consideration. This left me with five non-correlated weather variables: daily minimum and
5/19/2014 6/1/2014 6/21/2014 6/6/2014 7/18/2014 7/7/2014
Fig. 7. Maximum temperature increase weakly over the course of the season (r2=0.38) but remained highly variable from day to day.
26


maximum temperature, daily precipitation,
temperature range and precipitation on the prior day, and average rainfall over the preceding three days.
Objective A: Niche Delineation
Vegetative features identified as different for at least one species included: crown cover abundance of midgrass and forbs; basal cover abundance of midgrass, annual grasses, dung; both mean low and mean high VOR; and distance to the nearest shrub. I omitted subshrub and shrub cover from this comparative analysis, given that only Lark Bunting nests were associated with these features. Nest-sites of Horned Lark and McCown’s Longspur were more similar to each other than either was to Lark Bunting in
Crown Cover bv Miderass
MCLO HOLA LARB
Crown Cover bv Forbs
MCLO HOLA LARB
o
â– 'T
O
CO
O
o
Crown Cover bv Annual Grass
o
o
o
o
o
o
MCLO HOLA LARB
the case of several features (Fig. 8): crown abundance of midgrass and forbs; basal abundance of midgrass and annual grass; and mean high VOR. Interestingly, basal cover by
Fig. 8. Magnitude and direction of difference for vegetation features identified as having unequal distributions. For several features, vegetation was more similar between McCown’s Longspur (MCLO) and Homed Lark (HOLA) than between either of those species and Lark Bunting (LARB). For metrics of tall vegetation such as crown cover abundance of midgrass and forbs, cover was higher in Lark Bunting than in the other two species. Basal cover of annual grass showed the opposite trend, reflecting the higher abundance of sixweeks fescue in areas low in heavily grazed areas that are more likely to attract McCown’s Longspur and Homed Lark than Lark Bunting.
27


dung had equivalent similarity between Lark Bunting and McCown’s Longpur (0.78) and between Horned Lark and McCown’s Longspur
(p=0.77) but had low similarity between Lark Bunting and Homed Lark (0.22). Mean low VOR was more similar between McCown’s Longspur and Lark Bunting than between either of the other two pairs.
Table 3. Fit for time-based models.
Lark Bunting
Model # of Parameters AlCc AAlCc weight
Time + Nest Age 3 532.9 — 1.00
time + Nest Age 68 588.5 64.7 0.00
Global Model 69 590.7 66.8 0.00
Nest Age 2 653.8 129.9 0.00
Intercept Only 1 633.7 139.8 0.00
Time 2 665.7 141.8 0.00
time 67 718.2 194.3 0.00
time + Time 68 720.4 196.5 0.00
Horned Lark
Time + Nest Age 3 56.8 — 0.64
Intercept Only 1 59.1 2.39 0.19
Nest Age 2 60.4 3.7 0.10
Time 2 61.2 4.4 0.07
time + Nest Age 52 184.9 128.1 0.00
Global Model 53 190.1 133.4 0.00
time 51 191.9 135.4 0.00
time + Time 52 197.0 140.2 0.00
McCown's Longspur
Time + Nest Age 3 57.3 — 0.59
Time 2 58.7 1.5 0.28
Intercept Only 1 61.3 4.0 0.08
Nest Age 2 62.5 5.2 0.04
time + Nest Age 68 182.4 125.2 0.00
Global Model 69 187.8 130.5 0.00
time 67 191.3 134.0 0.00
time + Time 68 196.7 139.4 0.00
Time: a constant survival trend over time time: a factor by which survival varies from day to day Nest Age: days since a nest effort was initiated Global Model: including all possible parameters
28


Objective B: Drivers of Nest Survival
For all species, I found that a combination of time as a constant effect and nest age most strongly explained nest survival (Table 3). This relationship was high in all species, with little support for other models. According to the best model in each species, nest survival decreased with exposure time (LARB: P=-0.28, CI=-0.33:-0.22; HOLA: P=-0.22, CI=-0.35> 0.08; MCLO: P=-0.19, CI=-0.31:-0.08), but nest survival increased with nest age (LARB: P=0.25, CI=0.20:0.29; HOLA: P=0.21, CI=0.09:0.33; MCLO: (P=0.19, CI=0.08:0.29).
For Lark Bunting, a model including time, nest age, and daily precipitation was best supported by the data, followed distantly by the global model, and supported 2.5 times more than the global model (Table 4). According to the best-fitting model, daily precipitation had a negative impact on nest survival (P=-0.08, CI=-0.1 L-0.04). Relationships were less clear for Horned Lark and McCown’s Longspur, due to their small sample sizes. For Homed Lark, I again found that precipitation was a modest predictor of nest fate. This was followed by time and age only, average rain over the past three days, and temperature range on the preceding day. Precipitation on the last day had a mostly negative but inconsistent impact on nest survival (P=-0.13, CI=-0.27:0.00), while average rain over the past three days had a slight positive but inconsistent impact (P=0.13, CI=-0.14:0.39). Temperature range on the previous day had a moderately negative but inconsistent impact (P=-0.10, CI=-0.28:0.07). Due to the small sample size and strong risk of spurious results from comparing too many models, I selected a single best weather model by comparing existing models against one in which all top variables were included. This new model did not outperform the rain-only model. The relative weight in favor of the latter was 1.4 times that of the more complex model.
29


For McCown’s Longspur, variables related to rainfall occupied the three best-fitting models: average rain over the past three days, rain on the prior day, and rain on the last day. This was followed by the time only model and the global model. Support was low (AAICc>2) for temperature-related models.
I again compared existing models against one in which all top variables were included, and that new model was best supported.
Table 4. Fit for weather-based models. All Models include Time and Nest Age.
Lark Bunting
Model # of Parameters AICc AAICc weight
Rain 4 511.3 — - 0.70
Global Model 9 513.2 1.85 0.28
Min 4 518.5 7.18 0.19
Rainl 4 523.6 12.32 0.00
No Weather 3 523.9 12.53 0.00
Max 4 525.3 14.02 0.00
Rain3 4 525.5 14.15 0.00
Rangel 4 525.5 14.16 0.00
Intercept Only 1 663.7 152.4 0.00
Horned Lark
Min 4 56.7 — 0.17
No Weather 3 56.8 0.48 0.16
Rangel 4 56.8 0.51 0.16
Rain3 4 57.1 0.35 0.14
Rain 4 57.9 1.18 0.09
Max 4 58.6 1.84 0.07
Rainl 4 58.8 1.03 0.06
Intercept Only 1 59.1 2.43 0.05
Rain + Rain3 + Range 6 59.2 2.50 0.05
Global Model 9 59.4 2.64 0.04
McCown’s Longspur
Rain + Rain3 + Rainl 6 53.3 — 0.46
Rain 4 54.8 1.57 0.21
Rain3 4 56.4 3.15 0.09
No Weather 3 57.3 4.00 0.06
Rainl 4 58.0 4.78 0.04
Max 4 58.3 5.00 0.04
Min 4 58.3 5.01 0.04
Global Model 9 58.4 5.15 0.03
Range 4 59.4 6.14 0.02
Intercept Only 1 61.3 7.99 0.01
30


Rain: Total precipitation on the final day of the nest effort Rainl: Total precipitation on the penultimate day of the nest effort Rain3: Total Rain over the three days preceding the end of the nesting effort Min: Minimum temperature on the final day of the nesting effort Max: Maximum temperature on the final day of the nesting effort Range: Temperature range (max-min) on the final day of the nesting effort No Weather: Time and Nest Age, with no additional parameters
Global Model: including all possible parameters_______________________________________________
Precipitation on the last day had a negative impact on nest survival (P=-0.43, CI=-0.73>
0.12), precipitation on the penultimate day had a usually positive but inconsistent impact on
survival (P=2.02, CI=-0.34:4.39), and average rain over the past three days had a negative
impact on survival (P=0.19, CI=-0.33:-0.06).
My sample of nests where vegetation had been measured included 152 nests: 119 Lark Bunting nests, 17 Horned Lark nests, and 19 McCown’s Longspur nests. For Lark Bunting, I compared the fit of vegetation-based models with and without rain. Among the models including vegetation, five had competing support from the data: basal cover by bare ground (w=0.12), crown cover by dead vegetation (AAICc=0.37, w=0.10), crown cover by shrubs (AAICc=0.93, w=0.08), crown cover by forbs (AAICc=1.20, w=0.07), and crown cover by cacti (AAICc=1.30, w=0.07) All other models had poor support (AAICc>2), including daily rainfall. Nest survival showed a slight negative but inconsistent decrease with extent of bare ground (P=-0.02, CI=-0.04:0.01), crown cover by dead vegetation (P=-0.02, CI=-0.04:0.00), and crown cover by shrubs (P=-0.02, CI=-0.05:0.01). Nest survival showed a slight positive but inconsistent increase with crown cover by forbs (P=0.01, CI=0.00:0.02). For Homed Lark, I again compared the fit of vegetation-based models with and without rain. Due to the small sample size, I only built models including crown cover by midgrass, forbs, dead vegetation, and annual grasses; mean low and high VOR; and basal cover by dead vegetation and bare ground. In this case, the global model received the most support (w~l), indicating no clear superiority among models. This is likely a spurious result and signifies a
31


need for a larger sample size. For McCown’s Longspur, I compared the rain-based model against the same set of vegetation-based models included for Homed Lark. I found that the rain-based model received the most support (w=0.50), while a model including crown cover
Table 5. Fit for vegetation-based models. All models include Time and Nest Age.
Lark Bunting
Model # of Parameters AICc AAICc weight
BC Bare Ground 4 324.8 — 0.12
CC Dead Vegetation 4 325.1 0.37 0.10
CC Shrub 4 325.7 0.92 0.08
CC Forb 4 326.0 1.20 0.07
CC Cactus 4 326.0 1.26 0.05
Rain + BC Bare Ground 5 326.6 1.82 0.04
Rain + CC Cactus 5 327.0 2.23 0.04
CC Midgrass 4 327.0 2.27 0.03
BC Dead Veg. 4 327.3 2.51 0.03
CC Annual Grass 4 327.3 2.54 0.03
Rain + CC Shrub 5 327.5 2.76 0.03
Mean Hight VOR 4 327.6 2.85 0.03
Rain 4 327.7 2.89 0.03
Rain + CC Forbs 5 327.8 2.94 0.03
CC Subshrubs 4 327.7 2.94 0.03
Rain + CC Cactus 5 327.8 3.01 0.03
BC Dung 4 328.0 3.18 0.02
Mean Low VOR 4 328.0 3.19 0.02
CC Shortgrass 4 328.0 3.19 0.02
Rain + CC Midgrass 5 328.8 3.97 0.02
Rain + BC Dead Veg. 5 329.0 4.20 0.01
Rain + CC Annual 5 329.1 4.31 0.01
Rain + Mean High VOR 5 329.3 4.54 0.01
Rain + CC Shubshrub 5 329.4 4.65 0.01
Rain + Shrub Distance 5 329.5 4.70 0.01
Rain + BC Dung 5 329.5 4.70 0.01
Rain + Mean Low VOR 5 329.7 4.88 0.01
Rain + CC Shortgrass 5 329.7 4.89 0.01
Global Model 14 329.7 11.16 0.00
Intercept Only 1 409.6 84.82 0.00
Homed Lark
Global Model 12 37.9 — 1.00
CC Annual Grass 4 56.2 18.2 0.00
Mean Low VOR 4 57.3 19.4 0.00
Rain 4 57.9 20.0 0.00
BC Bare Ground 4 58.4 20.5 0.00
CC Midgrass 4 58.6 20.6 0.00
BC Dead Vegetation 4 58.6 20.7 0.00
CC Forbs 4 58.8 20.9 0.00
CC Dead Vegetation 4 58.8 20.9 0.00
Mean High VOR 4 58.8 20.9 0.00
Intercept Only 1 59.1 21.2 0.00
32


McCown’s Longspur
Rain + Rainl + Rain3 6 53.3 — 0.51
CC Dead Vegetation 4 54.8 1.55 0.23
CC Midgrass 4 57.0 3.77 0.08
BC Bare Ground 4 58.0 4.74 0.05
Mean Low VOR 4 59.1 5.83 0.03
CC Annual Grass 4 59.3 6.05 0.02
Table 5. (Continued)
Mean High VOR 4 59.4 6.01 0.02
CC Forbs 4 59.4 6.10 0.02
BC Dead Vegetation 4 59.4 6.14 0.02
Intercept Only 1 61.3 0.01 0.01
Global Model 14 67.8 0.00 0.00
BC: Basal Cover CC: Crown Cover
VOR: Visual Obstruction Reading Rain: Total precipitation on the final day of the nest effort Rainl: Total precipitation on the penultimate day of the nest effort Rain3: Total Rain over the three days preceding the end of the nesting effort
Min: Minimum temperature on the final day of the nesting effort Max: Maximum temperature on the final day of the nesting effort
Range: Temperature range (max-min) on the final day of the nesting effort Global Model: including all possible parameters
by dead vegetation received second-most support (AAICc=1.5. w=0.23). Support for other models was low (AAICc>2). In the best model, rain on the final day was indicated as having
a very strong and unambiguous positive impact on survival (P=-0.79 +/-0), which is
undoubtedly a spurious result, calling into question the rest of the impact estimates.
Discussion
I found that, as expected, nests of Horned Lark and McCown’s Longspur are more similar to one another than they are to nests of Lark Bunting, supporting the contention that the first two species nest primarily in shortgrass while Lark Bunting nests in taller grass. Contrary to expectation, I found that Homed Lark has lower niche breadth in most respects than Lark Bunting, suggesting that Lark Bunting is in fact a generalist while Horned Lark is more of a specialist. This indicates that, in addition to being a probable nomad, Lark Bunting gambles regarding nest-site vegetation, assuming greater risk at an individual level but perhaps
33


buffering the risk of population decline. In contrast, Horned Lark and McCown’s Longspur occupy narrow niches with respect to vegetation, indicating a lower assumption of risk at the individual level but a greater risk of population decline under sudden environmental change.
34


CHAPTER II
WEATHER RADAR DATA CORRELATE TO HAIL-INDUCED MORTALITY IN
GRASSLAND BIRDS Abstract
Hail contributes to avian mortality. Global climate change will increase the frequency of hail events, which may adversely influence avian population trends. However, estimating the magnitude of hail’s contribution to mortality or population demographics is challenging. Hail events are difficult to predict, and they often occur in locations where birds are not under scientific observation. Estimates of bird mortality through remote sensing would be useful for population monitoring, but observations of bird mortality are usually too imprecise to connect directly with spatial information on storm intensity. Here, we demonstrate a strong connection between Doppler weather reflectivity and nest fate (n=204) during an extreme hail storm that intercepted our study area on 22 June 2014. We provide strong evidence that high values of Doppler reflectivity correspond to hail-induced mortality in grassland birds, and that it can thus be used to estimate mortality at multiple scales. This event was part of a larger, ongoing study on grassland bird nest survival at the Central Plains Experimental Range (Weld Co., CO). The 2014 hail storm resulted in high but variably-distributed mortality among passerines. We attributed the spatial pattern of mortality to a heterogeneous distribution of hail sizes, and we hypothesized that—by serving as a proxy for hail size— Doppler reflectivity would accurately explain mortality distribution. To test this, we compared the spatial distribution of nest mortality to the spatial distribution of Doppler weather reflectivity at the time of the event. Average reflectivity was five decibels higher for locations where nests failed than for locations where nests survived; a threshold value of 62.5
35


dBZ correctly predicted 96% of nest fates. We conclude that Doppler reflectivity can be used to remotely estimate hail-induced nest mortality in populations of grassland birds. This will become an increasingly important tool as hail frequency grows under global climate change. Key Words Doppler data; severe weather; hail; grassland birds; nest mortality..
Introduction
Severe weather can kill birds (Wauer and Wunderle 1992; Wiley and Wunderle 1993; Newton 2007; Frederiksen et al. 2008; McKechnie and Wolf 2010; Saunders et al. 2011; McKechnie et al. 2012), but quantifying weather’s contribution to mortality is challenging. In the past, severe weather may have had a negligible long-term impact on bird populations, because bird species evolved in the context of weather events. However, under global climate change, the frequency and intensity of severe weather are increasing (Fischer and Knutti 2015), potentially exceeding species’ ability to compensate for and recuperate from these events.
Hail is a common severe weather phenomenon in North America (Changnon 2008), and it can cause high bird mortality in areas where individuals are concentrated (Diehl et al.
2014). A species whose range is limited to a small, hail-prone area can experience a significant population decline due to even a single major event (Saunders et al. 2011). Thus, we can expect for hail to disproportionately affect species that breed in habitat islands, where adults and young are concentrated within a small area and the surrounding habitat matrix serves as a poor population reservoir (Duelli and Obrist 2003).
The North American shortgrass steppe is an example of a habitat type reduced to noncontiguous islands because of historical land use and habitat heterogeneity. This region is regularly subject to hail (Fig. 7). This habitat type occurs along the semi-arid western edge
36


of the Great Plains, where a combination of low annual precipitation and high evaporative potential (Sala et al. 1992) result in a plant community dominated by low-stature grasses (Milchunas et al. 1989). Several bird species breed primarily on the shortgrass steppe, and nests of most are constructed on or near the ground with little vegetative cover (Knopf 1996), making them vulnerable to severe weather. Based on climate, the potential range of the shortgrass steppe runs from northern New Mexico and Texas to southern Saskatchewan, but habitat across much of this range has been converted to agriculture (Wylie et al. 2002) or altered through the disruption of the native grazing regime (Demer et al. 2009). The
37


shortgrass steppe is now limited to islands of semi-natural habitat such as the Pawnee National Grassland in northern Colorado.
The Pawnee National Grassland is located in the North American hail belt, a broad region where hail occurs with higher frequency than in other parts of the continent (Allen et al. 2015). Thus, hail is not an abnormal contributor to mortality for the birds that breed there. However, the frequency of damaging hail storms has increased, and many climate models suggest that this trend will continue into the foreseeable future (Trapp et al. 2007; Kapsch et al. 2012; Paquin et al. 2014; Allen et al. 2015) (e.g., Trapp et al. 2007, Kapsch et al. 2012, Paquin et al. 2014, Allen et al. 2015). At least one model, however, predicts a decrease in the frequency of damaging hail over Colorado as the surface warms (Mahoney et al. 2012). This highlights the uncertainty associated with predicting future climate but underscores the need for an accurate metric to remotely estimate hail impact as climate change progresses.
The ability to estimate hail-induced mortality may mean the difference between effective and null management of grassland bird habitat. This is challenging because storm location and intensity are heterogeneously distributed (Morgan and Towery 1975) and are not easily predicted (Clark et al. 2012). Thus, studies cannot be designed to evaluate the impacts of naturally occurring extreme weather events; frequently the zone of highest intensity within a weather system may occur where birds are not under scientific observation. Quantifying the impact of hail on bird populations requires the right remote-sensing technology that can be applied at a large spatial scale and can capture local variability in storm location and intensity. The metrics produced must distinguish between rain and hail, as well as between small and large hail, as these different hydrometeor classes may cause very different outcomes for breeding birds. Here, we introduce Doppler weather radar as a tool for remotely
38


capturing high-resolution, large-scale information about storms, and we present a case example demonstrating the connection between Doppler reflectivity and hail-induced mortality in breeding birds in a shortgrass steppe habitat island.
The United States Doppler weather radar network provides high-coverage data on atmospheric conditions. Stationary active sensors emit pulses of electromagnetic radiation at multiple angles above the ground, and the sensors then measure the intensity of returning radiation (reflectivity) in decibels relative to a standard raindrop (dBZ). Reflectivity is positively correlated with the density of precipitation, and it can be used to reliably distinguish between pure rain and rain mixed with hail (Kunz and Kugel 2015). Reflectivity increases with hail diameter (Hohl et al. 2002; Hazenberg et al. 2011). Doppler radar sensors emit radiation in a fixed beam-width measured in degrees. Resolution (number of unique reflectivity values per unit area) has increased with the deployment of the Next Generation Radar (NEXRAD) network, which uses a narrower beam than the first-generation radars (Klazura and Imy 1993). Still, resolution decays with distance from the emitter, and this must be taken into account when estimating hail impact. Here, we demonstrate the application of NEXRAD data for estimating local storm intensity, as well as the use of station distance for quantifying spatial resolution.
This study was based on nest survival data gathered during the first year of a long-term nest monitoring project at the Central Plains Experimental Range (CPER), a shortgrass steppe grazing research facility adjacent to the Pawnee National Grassland. We analyzed the distribution of nest mortality and NEXRAD reflectivity after a severe hail storm in 2014 that had resulted in mass mortality of breeding birds. Our objective was to quantify the accuracy of reflectivity for explaining hail-induced nest mortality resulting from this single event. In
39


so doing, we assessed the value of this tool for remotely estimating the impact of hail on known populations of breeding grassland birds.
Methods
Study Area
The CPER is located in Weld County, Colorado, on the west unit of the Pawnee National Grassland. The climate is semi-arid, with most precipitation occurring in the summer months (Sala et al. 1992). Severe weather including extreme rain and hail is a common feature of summertime weather, but its distribution is heterogeneous and unpredictable (Rosenberg 1987). Vegetation at the CPER is dominated by low-stature grasses, with smaller proportions of cacti, midgrasses, and forbs (Lauenroth and Sala 1992). The woody plant community includes both tail-stature shrubs and low-stature subshrubs (Lee and Lauenroth 1994), which might be expected to afford protection from hail. Most bird species that breed at the CPER belong to the order Passeriformes. This includes Lark Bunting (Calamospiza melanocorys), Horned Lark (Eremophila alpestris), Western Meadowlark (Sturnella neglecta), McCown’s Longspur (Rhynchophanes mccownii), Grasshopper Sparrow (Ammodramus savannarum), and Brewer’s Sparrow (Spizella breweri). These are small-bodied organisms that should not be able to survive the physical impact of large hail. The dominant nesting mode at the site is a partially recessed cup in the ground with little or no woody vegetation over the nest. Ground nests are often placed beside an emergent clump of grass or cactus, which may afford protection from solar radiation, but not from wind or precipitation (With and Webb 1993). Very few nests are covered by the type of vegetation that would protect nests from the impact of hail.
40


Nest Site Measurements
From May-July 2014, we located nests by rope-dragging and completed high-frequency (2-4 day interval) nest checks, as part of a larger study examining the relationship between nest survival and livestock grazing regime. We monitored 204 nests that were active just prior to the hail storm, and analyses reported here pertain to that sample. This included 186 Lark Bunting, seven Horned Lark, three Mourning Dove, two McCown’s Longspur, two Grasshopper Sparrow, two Brewer’s Sparrow, and two Western Meadowlark nests. All 204 nests had been last checked as recently as 2-3 days prior to the hail storm, and all were checked again the morning after the storm. We measured nest-site vegetation as part of the larger study, but we were not able to measure vegetation at every nest. Due to logistical constraints, we only measured vegetation at 160 of the 204 hail-storm nests. This included 145 Lark Bunting nests, six Horned Lark, two McCown’s Longspur, two Grasshopper Sparrow, two Western Meadowlark, two Mourning Dove, and one Brewer’s Sparrow nests. Vegetation-based analyses were applied to that subset.
The storm occurred 22 June 2014, and a cell producing large hail crossed our study area along a northwest to southeast trajectory (-1800 - 1900 MST). Spotters reported hail up to 3.18 cm in diameter, and the volume of hail was estimated to be relatively high. Hail size and force were sufficient to break windows on buildings at the study area. When checking the fate of nests after the storm, we treated any failed nest as having failed due to the storm. In most cases, failed nests contained crushed eggs or dead chicks. At many nest sites, we also found a dead adult on or near the nest. A few of the nests were empty, but we treated these as having failed due to the storm and assumed that the contents had been scavenged during the several hours since the passing of the storm. We classified nests with live chicks, whole eggs,
41


and/or a live adult on the nest as still active, and we verified the survival of these nests through continued monitoring.
To quantify storm intensity in relation to our nest distribution, we obtained National Weather Service NEXRAD Base Reflectivity at the 0.5° elevation angle for 22 June 2014 from NOAA Climate Data Online (https://www.ncdc.noaa.gov/cdo-web/). The data were collected at the National Weather Service station in Cheyenne, Wyoming. Distance from the weather station to nests in the study area ranged from 27 to 37 km. Doppler beam width was 0.925°. In order to quantify reflectivity resolution, we calculated both mean distance between nests and reflectivity grid cell width. To achieve the former, we imported nest coordinates to QGIS (version 2.8.1-Wien) and calculated mean distance between nests. To achieve the latter, we estimated reflectivity value grid cell width as the length of an arc with an inside angle equal to the beam width and a radius equal to distance from the station. We calculated reflectivity resolution as the number of grid cells (and hence unique values) per square kilometer. In order to do so, we quantified the number of nests—based on mean distance between nests—that would fit inside a grid cell. We assumed that reflectivity would increase with hydrometeor size and density and that other atmospheric phenomena known to reflect Doppler radiation would not have readings exceeding those of the largest hail produced by the storm. For each nest site, we determined maximum reflectivity.
We hypothesized that terrain and vegetation would be important to nest survival. The terrain variables that we focused on were elevation, slope, and aspect, and we expected to find an interaction effect among terrain variables (Table 6). Elevation, slope, and aspect would all affect survival, but the effect of each one would depend on values for the others. For example, nests on a northwest-facing slope should have had lower survival than nests on
42


a southeast-facing slope or nests in the lee of a hill, given that the storm was traveling along a northwest to southeast trajectory. We obtained a 10-m resolution Digital Elevation Model (DEM) (http://coloradoview.org/cwis438/websites/ColoradoView/), and we used it to calculate slope and aspect for our study area. For each nest site, we determined elevation, slope, and aspect. We further hypothesized that nest-site vegetation would strongly influence nest survival. We expected woody vegetation to provide protection, and we quantified this as percent cover by shrubs or subshrubs. We assumed that no other vegetation class (i.e. shortgrass, midgrass, annual grass, forbs, cacti) would provide protection from hail.
Data Analysis
We used a three-pronged approach to quantifying the relationship between Doppler reflectivity and nest fate. First, we compared distributions of reflectivity values by fate using Kruskal-Wallis one-way analysis of variance (Kruskal and Wallis 1952). We assessed magnitude of difference between distributions using the H statistic and associated p-value. Second, we quantified the predictive accuracy of sequential reflectivity values at 0.5 dBZ intervals between the minimum and maximum reflectivity levels measured. For each value, we treated nests associated with equal or greater reflectivity as having failed, and we treated nests associated with lower reflectivity as having survived. We calculated accuracy as overlap between expected and observed numbers of nests in each fate category. In order to assess magnitude of difference between expected and observed fate outcomes at each reflectivity level, we used a Chi-square goodness of fit test.
43


Table 6. Model Hypotheses and Structure. We based model construction on a priori hypotheses regarding the relationship between nest survival and reflectivity (dbz), elevation (ele), slope (sip), and aspect (asp).
Hypothesis Nest survival... Structure
declines with increasing reflectivity, as large hail is more likely to kill birds. dbz
increases with increasing elevation, as low-lying areas are prone to flooding. ele
is highest for nests facing away from rather than into the storm. asp
is highest for high-elevation nests facing away from the storm. ele*asp
is highest nests on steep slopes facing away from the storm. slp*asp
is highest for high-elevation nests on steep slopes facing away from the storm. ele*slp*asp
increases with increasing elevation but decreases with increasing reflectivity. ele+dbz
is highest for nests facing away from the storm in low-reflectivity areas. asp+dbz
is highest for high-elevation nests facing away from the storm in low-reflectivity areas. ele*asp+dbz
is highest for nests on steep slopes facing away from the storm in low-reflectivity areas. slp*asp+dbz
is highest for high-elevation nests facing away from the storm on steep slopes in low-reflectivity areas. ele*slp*asp+ dbz
Our third approach to analyzing the link between reflectivity and fate was the
comparison of fit for generalized linear models based on a priori hypotheses regarding the relationship between nest fate and nest site characteristics. We fit two separate sets of models: a terrain- and reflectivity-based set for the full sample of 204 nests; and a terrain-, reflectivity-, and vegetation-based set for the subset of 160 nests where vegetation had been measured. The first set included 11 models—as well as an intercept-only and a global model (Table 1). The second set included five models: one each for shrub cover and subshrub cover, another two for shrub or subshrub cover with the best model from the first set, and the global and intercept-only. We fit models using R (version 3.2.2) package RMark (Laake and Rexstad 2014), which uses the R language interface to run program MARK (White and Burnham 1999). We evaluated model fit using AAICc, and we assessed data support for models using model weight (w). We quantified relative support for each model using the relative weight in favor of that model, compared to the best-supported model. We assessed a variable’s effect on nest fate using slope estimate (P) and confidence interval (Cl), noting whether the estimate was negative or positive and the extent to which the Cl overlapped zero.
44


Results
Mortality and reflectivity were heterogeneously distributed across the study area (Fig. 8). Reflectivity was on average five decibels higher for nests that failed during the storm than for nests that survived (77=62.18,/KO.001) (Fig. 9). Mean distance between nests was 95.9 m. Reflectivity resolution for the site ranged from 2.27 to 1.67 grid cells per square kilometer. Thus, the number of nests that would be assigned the same value based on distance from the station ranged from 4.6 to 6.2. Explanatory accuracy was highest (96.1%) for a reflectivity of 62.5 dBZ (Fig. 10), and the difference between expected and observed nest outcomes based on that threshold was minimal (x2=0.08,/>=0.22). Among the reflectivity- and terrain-based models, the model including reflectivity and elevation was
Fate
o Survived A Failed
dBZ
45 - 50 50 - 55 55-60
60 - 65 65 - 70 70- 75
0 2 4 6 km
Figure 8. Distribution of nest fates and Doppler reflectivity across the study area. More nests failed in the southwestern half of the study area than in the northeastern half. Reflectivity was higher in the southwestern half of the study area than in the northeastern half
45


most supported by the data (w=0.36), followed by reflectivity alone (AAICc=l. 11, w=0.21) (Table 7). Thus, the best model was 1.71 times more supported by the data than the second-best model. Other models were not well supported by the data. Based on the best model, the effect of reflectivity was strongly negative (P=-0.24, CI=[-0.31:-0.17]. The effect of elevation was more neutral (P=-0.12, CI=[-0.03:0.00]. Within the set of models including cover by shrubs and subshrubs, the reflectivity- and elevation-based model was still best supported (w=0.49), followed by a combination of reflectivity, elevation, and cover by shrubs (AAICc=1.49, w=0.23). The best model from this set was 2.13 times more supported by the data than the second-best model. According to the best model, the effect of reflectivity was consistently negative (P=-0.24, CI=[-0.31:-0.17], but the effect of elevation was neutral (P=0.12, CI=[-0.03:0.00].
Survived Failed
Figure 9. Distribution of reflectivity by fate. Nests that failed were associated higher and more consistent reflectivity than nests that survived.
46


Table 7. Results of model fitting. We fit terrain-based models first and then used the best model from that set in subsequent vegetation-based models. These were based on reflectivity (dbz), elevation (ele), slope (sip), aspect (asp), crown cover by shrubs (cshr), and crown cover by shubshrubs (csub). Vegetation-based models included both models with the best terrain parameters identified from the first set, and models without. Here, only models with AAICc<2 are shown. Support for other models was negligible.
Model AAICc weight
Terrain-based Models
dbz+ele — 0.35
dbz 1.11 0.20
dbz+slp*asp 1.54 0.17
dbz+ele* asp 1.86 0.14
Vege tation-based Models
dbz+ele — 0.49
dbz+ele+cshr 1.48 0.23
dbz+ele+csub 1.89 0.19
Figure 10. Explanatory accuracy of reflectivity thresholds. Where all nests with a value below a given level are predicted to have survived, while all of those with a value at or above that level are predicted to have failed. Predicted numbers of failed and surviving nests were closest to observed for a reflectivity value of 62.5 dBZ. Accuracy was 96.1%.
47


Discussion
High Doppler reflectivity values are a strong indicator of hail-induced mortality in grassland birds. This conclusion is based on three pieces of evidence: reflectivity distribution, accuracy of the best reflectivity threshold, and weight in favor of a reflectivity-based nest survival model. We therefore assert that reflectivity can be used to remotely estimate hail-induced mortality in known populations of grassland birds that may not be under direct observation. Reflectivity may be useful for predicting hail-induced mortality in birds breeding in other habitat types, with the caveat that vegetation and terrain may matter more in other habitats. The impact for birds nesting in more complex nest microhabitat will likely be more strongly tied to vegetation and terrain. Nevertheless, reflectivity should be a strong predictor of hail-induced mortality in most open-cup nesters.
We found that reflectivity was significantly higher for nests that failed than for nests that survived. Thunderstorm reflectivity is strongly correlated with precipitation rate (Hazenberg et al. 2011), and levels greater than 55 dBZ indicate hail (Schiesser 1990). We assumed that all nest failures associated with this event were a result of direct impact by large hail, and we hypothesized that an increase in reflectivity would be linked to an increase in mortality. A weak connection between reflectivity and fate could have been interpreted either as attributing too many failures to hail when they were due to other causes; or poor capacity for reflectivity alone to predict outcome. Because we found a strong relationship between reflectivity and nest outcome, we concluded that our assumption regarding the agency of mortality was correct for most nests and that our hypothesis regarding the link between reflectivity and fate was supported.
48


We found that a reflectivity value of 65 dBZ was the tipping point for mortality and significantly explained nest fate. Despite numerous other storms that year involving high-volume rainfall and small hail, no other storm caused direct mortality in nests under observation. Previous studies suggest a rainfall-mediated relationship between predation and nest mortality (e.g., Lehman et al. 2008), but evidence for direct mortality via drowning or nest abandonment appears limited to birds nesting on the ground near waterbodies (e.g., Sexson and Farley 2012). Reports have established that hail can kill birds (Hall and Harvey 2007; Newton 2007) (Hall and Harvey 2007, Newton 2007, Diehl et al. 2014), but to our knowledge no other study has quantified the lower size limit for hail that is fatal. The accuracy of our mortality threshold (96%) supports our anecdotal observation that hail size matters. Based on these results, we conclude that small hail associated with reflectivity in the 55-65 dBZ range is not fatal to small passerines, so hail presence alone is insufficient to explain fate, and size must be taken into account. Nest habitat variables did not explain variation in nest fate. While vegetation may mediate the impact of solar radiation in shortgrass birds (With and Webb 1993), and while woody vegetation appears to play an important role in predation risk (With 1994), its contribution to explaining the distribution of hail-induced mortality in this storm was negligible. Likewise, terrain turned out to be unimportant. Although elevation appeared in the top models from both candidate sets, its influence was neutral, indicating that this may be a spurious artifact. We included habitat variables in the set of candidate survival models in an effort to account for unexplained variation in nest fate. We thought that shrubs might provide protection from hail, that low-lying nests might be more prone to flooding and hail accumulation, and that nests on leeward steep slopes might be buffered from hail impact. However, none of these variables had a
49


consistent association with nest fate, and we concluded that they were not important mediators of hail vulnerability in this system. The poor fit of terrain-based models to this dataset does not necessarily indicate that terrain is not important to nest survival during hail storms. For example, a cluster of nests that survived in a high-reflectivity area may have been protected because they were in the lee of a tall hill (Figure 4). The effect of such topographic features may be poorly quantified by simple measurements such as elevation, terrain, and aspect. Accounting for the influence of complex terrain may require more comprehensive metrics of proximity to rugged terrain. The survival of nests with more complex nest structure or on more rugged terrain might have a stronger dependence on habitat, in which case the contribution of these features to survival should be quantified. We expect that accurate reflectivity-based mortality estimates can be derived for most other open-cup ground-nesting birds.
We attribute unexplained variability in nest fate to distance-based degradation in reflectivity resolution. United States Doppler weather technology and coverage have improved substantially over the past 60 years (Brown and Lewis 2005) (Klazura and Imy 1993, Brown and Lewis 2005). We can now combine reflectivity with other measurements to more accurately distinguish between rain and hail, as well as estimate hail size (Kunz and Kugel 2015). Fine-scale resolution means that we can also readily distinguish the characteristics of different parts of a storm cell (Hardegree et al. 2008). However, the number of unique values obtainable per unit area is still contingent on distance from the transmitter. Hence, the large size of reflectivity grid cells relative to the mean distance between nests de facto resulted in some inaccurate assignments. Tobler’s First Law of Geography (Tobler 1970), stating that similarity between objects increases with spatial proximity, holds true for
50


many ecological phenomena (Miller 2004) but appears poorly suited to the distribution of precipitation within a storm. Therefore, space-based binning of nests may have resulted in a large difference between estimated and true hail size for some nests.
Hail is important because it may add to mortality rates induced by other factors, lowering population viability. A single event may decrease annual production by killing both young and adults. The 2014 hail storm was the primary cause of nest failure that year and resulted in lower estimated nest survival than in 2015, a year that was free of major storms. All species covered by this study have the capacity to re-nest, so nests destroyed by hail may be replaced by surviving adults within the same season. Conceivably, release from competitive reproductive restraint may lead surviving adults to lay bigger clutches, stabilizing productivity for the year (Both 1998). However, even in the absence of additional storms,
.3?
Fig. 11. A survival anomaly likely related to terrain. This cluster of twelve nests should have all failed based on local reflectivity. Vegetation data were absent for most of these nests. Survival may have been due to cover by woody vegetation. Alternatively, the apparent contradiction may be an artifact of reflectivity resolution. A third possible explanation is that those nest were protected because they were in the lee of a hill with the highest prominence in the study area. Quantifying topographic context is challenging
51


probability against success is moderate to high due to the frequency of nest predation in grassland ecosystems (e.g., Vickery et al. 1992; Pietz and Granfors 2000; Murray 2015). Multiple severe hail storms within the same season could negate individuals’ ability to renest successfully. An alternative possibility is that adults killed by hail may be replaced by individuals immigrating from adjacent areas, restoring the local population size. Yet, this is contingent on proximity to suitable habitat with a reservoir population, and in many cases grassland habitat islands are too isolated for this to be expected.
The study also demonstrates the strong link that can exist between remotely sensed weather data and directly observed mortality patterns, promoting the value of Doppler data for estimating the impact of hail events on nesting birds across a larger area than that typically encompassed by direct monitoring efforts. Doppler data may serve as an important tool in monitoring bird populations. The frequency of large-diameter hail storms has increased due to global climate change, and it is projected to continue increasing. Doppler data make it possible to estimate the extent of large-diameter hail impact on nesting birds. Combined with known ranges of breeding birds, it should be possible to estimate hail-induced mortality at the regional scale. Further, based on climate models and the expected increase in hail frequency, adding hail to the population projection should allow us to generate more accurate estimates for future population sizes given no change in other management parameters. Using this information, we can identify changes to management aspects that will most benefit bird populations. It is likely to emerge that increasing contiguity among protected areas will be most beneficial to breeding birds, because that will provide a means to compensate for the effects of high-frequency hail events.
52


LITERATURE CITED
Adams CC. 1908. The Ecological Succession of Birds Published by : American
Ornithologists ’ Union Stable URL : http://www.jstor.org/stable/4070690 All use subject to http://about.jstor.org/terms THE AUK : Auk 25:109-153.
Aguilar TM, Dias RI, Oliveira AC, Macedo RH. 2008. Nest-site selection by Blue-black Grassquits in a Neotropical savanna: Do choices influence nest success? J. F. Ornithol. 79:24-31.
Akaike H. 1974. A new look at the statistical model identification. IEE Trans. Autom. Control 19:716-723.
Allen JT, Tippett MK, Sobel AH. 2015. An empirical model relating U.S. monthly hail occurrence to large-scale meteorological environment. J. Adv. Model. Earth Syst. 7:226-243.
Augustine DJ, Baker BW. 2013. Associations of grassland bird communities with blacktailed prairie d in the North American Great Plains. Conserv. Biol. 27:324-334.
Augustine DJ, Demer JD. 2015. Patch-burn grazing management, vegetation heterogeneity, and avian responses in a semi-arid grassland. J. Wildl. Manage. 79:927-936.
Beason RC, Franks EC. 1974. Breeding behavior of the Horned Lark. Auk 91:65-74.
Beja P, Schindler S, Santana J, Porto M, Morgado R, Moreira F, Pita R, Mira A, Reino L. 2014. Predators and livestock reduce bird nest survival in intensive Mediterranean farmland. Eur. J. Wildl. Res. 60:249-258.
Benson TJ, Chiavacci SJ, Ward MP. 2013. Patch size and edge proximity are useful
predictors of brood parasitism but not nest survival of grassland birds Author ( s ): Thomas J . Benson , Scott J . Chiavacci and Michael P . Ward Published by : Wiley
53


Stable URL : http://www.jstor.org/stable/23440. Ecol. Appl. 23:879-887.
Blank PJ, Dively GP, Gill DE, Rewa CA. 2011. Bird community response to filter strips in Maryland. J. Wildl. Manage. 75:116-125.
Bleho BI, Koper N, Machtans CS. 2014. Direct effects of cattle on grassland birds in Canada. Conserv. Biol. 28:724-734.
Bogard HJK, Davis SK. 2014. Grassland songbirds exhibit variable responses to the proximity and density of natural gas wells. J. Wildl. Manage. 78:471-482.
Bollinger EK, Gavin TA. 2004. Responses of nesting Bobolinks (Dolichonyx oryzivorus) to habitat edges. Auk 121:767-776.
Both C. 1998. Density dependence of clutch size: habitat heterogeneity or individual adjustment? J. Anim. Ecol. 67:659-666.
Brown R a., Lewis JM. 2005. Path to NEXRAD: doppler radar development at the National Severe Storms Laboratory. Bull. Am. Meteorol. Soc. 86:1459-1470.
Camfield A, Martin K. 2009. The influence of ambient temperature on horned lark incubation behaviour in an alpine environment. Behaviour 146:1615-1633.
Camfield AF, Clason AJ, Martin K. 2007. Mother-son parental care in Homed Larks. Wilson J. Ornithol. 119:302-304.
Canfield RH. 1941. Application of the line interception method in sampling range vegetation. J. For. 39:388-394.
Cao J, He CZ, Wells KMS, Millspaugh JJ, Ryan MR, Ryan MR, Cao J, He CZ, Wells KMS, Millspaugh JJ. 2009. Modeling Age and Nest-Specific Survival Using a Hierarchical Bayesian Approach. Int. Biometric Soc. 65:1052-1062.
Carson RJ, Spicer GS. 2003. A phylogenetic analysis of the emberizid sparrows based on
54


three mitochondrial genes. Mol. Phylogenet. Evol. 29:43-57.
Casas F, Vinuela J. 2010. Agricultural practices or game management: which is the key to improve red-legged partridge nesting success in agricultural landscapes? Environ. Conserv. 37:177-186.
Chaine AS, Lyon BE. 2008. Adaptive plasticity in female mate choice dampens sexual selection on male ornaments in the lark bunting. Science (80-. ). 319:459-462.
Chaiyarat R, Eiam-Ampai K. 2014. Effects of grazing on nesting of the oriental pratincole
(Glareola maldivarum) in Bueng Boraphet Wetland, Thailand. Zoolog. Sci. 31:77-82.
Changnon SA. 2008. Temporal and spatial distributions of wind storm damages in the United States. Clim. Change 94:473-482.
Churchwell RT, Davis C a., Fuhlendorf SD, Engle DM. 2008. Effects of patch-burn
management on Dickcissel nest success in a tallgrass prairie. J. Wildl. Manage. 72:1596-1604.
Clark AJ, Weiss SJ, Kain JS, Jirak IL, Coniglio M, Melick CJ, Siewert C, Sobash RA, Marsh PT, Dean AR, et al. 2012. An overview of the 2010 Hazardous Weather Testbed experimental forecast program spring experiment. Bull. Am. Meteorol. Soc. 93:55-74.
Cody ML. 1968. On the methods of resource division in grassland bird communities. Am. Nat. 102:107-147.
Davies SJJF. 1982. Behavioural adaptations of birds to environments where evaporation is high and water is in short supply. Comp. Biochem. Physiol. — Part A Physiol. 71:557-566.
Davis SK. 2004. Area sensitivity in grassland passerines : Effects of patch size , patch shape ,
55


and vegetation structure on bird abundance and occurrence in southern Saskatchewan. Auk 121:1130-1145.
Davis SK. 2005. Nest-Site Selection Patterns and the Influence of Vegetation on Nest Survival of Mixed-Grass Prairie Passerines. Condor 107:605-616.
Davis SK, Brigham RM, Shaffer, TL, James PC. 2006. Mixed-grass prairie passerines exhibit weak and variable responses to patch size. Auk 123:807-821.
Dean WRJ, Williams JB. 2004. Adaptations of birds for life in deserts with particular reference to Larks (Alaudidae). Trans. R. Soc. South Africa 59:79-91.
Degregorio BA, Chiavacci SJ, Weatherhead PJ, Willson JD, Benson TJ, Sperry JH. 2014. Snake predation on North American bird nests: Culprits, patterns and future directions. J. Avian Biol. 45:325-333.
Derner JD, Lauenroth WK, Stapp P, Augustine DJ. 2009. Livestock as ecosystem engineers for grassland bird habitat in the western Great Plains of North America. Rangel. Ecol. Manag. 62:111-118.
Diehl RH, Bates JM, Willard DE, Gnoske TP. 2014. Bird mortality during nocturnal migration over Lake Michigan: a case study. : 19-29.
Dinsmore SJ, White GC, Knopf FL. 2002. Advanced techniques for modeling avian nest survival. Ecology 83:3476-3488.
Dreitz VJ, Conrey RY, Skagen SK. 2012. Drought and cooler temperatures are associated with higher nest survival in Mountain Plovers.
Du B, Liu C-J, Bao S-J. 2015. Begging form and growth pattern of nestlings correlate with parental food-allocation patterns in the Horned Lark (Eremophila alpestris). Can. J. Zool. 93:273-279.
56


DuBois AD. 1937. The McCown’s Longspurs of a Montana Prairie. Condor 39:233-238.
Dubois BAD. 1935. DuBois, A Dawes. Condor 37:56-72.
Duelli P, Obrist MK. 2003. Regional biodiversity in an agricultural landscape: the contribution of seminatural habitat islands. Basic Appl. Ecol. 4:129-138.
Felske BE. 1971. The population dynamics and productivity of mccown’s longspur at matador, Saskatchewan. University of Saskatchewan.
Fischer EM, Knutti R. 2015. Anthropogenic contribution to global occurrence of heavy-precipitation and high-temperature extremes. Nat. Clim. Chang. 5:560-564.
Frederiksen M, Daunt F, Harris MP, Wanless S. 2008. The demographic impact of extreme events: stochastic weather drives survival and population dynamics in a long-lived seabird. J. Anim. Ecol. 77:1020-1029.
Giocomo JJ, Moss ED, Buehler DA, Minser WG. 2008. Nesting biology of grassland birds at Fort Campbell, Kentucky and Tennessee. Wilson Bull. 120:111-119.
Glover FA. 1953. Nesting ecology of the Pied-Billed Grebe in Northwestern Iowa. Wilson Bull. 65:32-39.
Glover FA. 1956. Nesting and production of the Blue-Winged Teal (Anas discors Linnaeus ) in Northwest Iowa. J. Wildl. Manage. 20:28-46.
Grant TA, Shaffer TL. 2012. Time-specific patterns of nest survival for ducks and passerines breeding in North Dakota. Auk 129:319-328.
Greer RD, Anderson SH. 1989. McCown’s Longspurs and Habitat Resources. Condor 91:609-619.
Grinnell J. 1917. The Niche-Relationships of the California Thrasher. Auk 34:427-433.
Hall DW, Harvey TM. 2007. Mortality at a Night Roost of Great-Tailed Grackles and
57


European Starlings during a Spring Hail Storm. Wilson J. Ornithol. 119:309-312.
Hardegree SP, Van Vactor SS, Levinson DH, Winstral AH. 2008. Evaluation of NEXRAD Radar Precipitation Products for Natural Resource Applications. Rangel. Ecol.
Manag. 61:346-353.
Hartman CA, Oring LW. 2009. Reproductive success of song-billed curlews (Numenius americanus) in Northeastern Nevada hay fields. Auk 126:420-430.
Hatchett ES, Hale AM, Bennett VJ, Karsten KB. 2013. Wind turbines do not negatively affect nest success in the Dickcissel (Spiza americana). Auk 130:520-528.
Hazenberg P, Yu N, Boudevillain B, Delrieu G, Uijlenhoet R. 2011. Scaling of raindrop size distributions and classification of radar reflectivity-rain rate relations in intense Mediterranean precipitation. J. Hydrol. 402:179-192.
Helzer CJ, Jelinski DE. 1999. The Relative Importance of Patch Area and Perimeter-Area Ratio to Grassland Breeding Birds. Ecol. Appl. 9:1448-1458.
Henderson AE, Davis SK. 2014. Rangeland health assessment: a useful tool for linking range management and grassland bird conservation? Rangel. Ecol. Manag. 67:88-98.
Henningsen JC, Best LB. 2005. Grassland bird use of riparian filter strips in Southeast Iowa. J. Wildl. Manage. 69:198-210.
Hohl R, Schiesser HH, Aller D. 2002. Hailfall: The relationship between radar-derived hail kinetic energy and hail damage to buildings. Atmos. Res. 63:177-207.
Holt RD, Keitt TH. 2005. Species’ borders: A unifying theme in ecology. Oikos 108:3-6.
Igl LD, Johnson DH. 1999. Le Conte’s sparrows breeding in conservation reserve program
fields: precipitation and patterns of population change. Stud. Avian Biol. 19:178-186.
Irvin E, Duren KR, Buler JJ, Jones W, Gonzon AT, Williams CK. 2013. A multi-scale
58


occupancy model for the Grasshopper Sparrow in the Mid-Atlantic. J. Wildl. Manage. 77:1564-1571.
Jehle G, Yackel Adams AA, Savidge JA, Skagen SK. 2004. Nest survival estimation: a review of alternatives to the Mayfield estimator. Condor 106:472.
Johnson DH, Grier JW. 1988. Determinants of breeding distribution of ducks. Lincoln, Nebraska, USA Report No.: 242.
Johnson TN, Kennedy PL, Etterson MA. 2012. Nest success and cause-specific nest failure of grassland passerines breeding in prairie grazed by livestock. J. Wildl. Manage. 76:1607-1616.
Jones SL, Dieni JS, Green MT, Gouse PJ, The S, Journal W, Mar N, Jones SL, Dieni JS, Green MT, et al. 2007. Annual Return Rates of Breeding Grassland Songbirds Published by : Wilson Ornithological Society Stable URL :
http://www.jstor.org/stable/20455951 REFERENCES Linked references are available on JSTOR for this article : ANNUAL RETURN RATES OF BREEDING GRASS. Wilson J. Omithol. 119:89-94.
Jongsomjit D, Jones SL, Gardali T, Geupel GR, Gouse PJ. 2007. A guide to nestling
development and aging in altricial passerines. Washington, D.C. Biological Technical Publication Report No.: FWS/BTP-R6008-2007.
Kapsch ML, Kunz M, Vitolo R, Economou T. 2012. Long-term trends of hail-related weather types in an ensemble of regional climate models using a Bayesian approach. J. Geophys. Res. Atmos. 117:1-16.
Kerns CK, Ryan MR, Murphy RK, Thompson FR, Rubin CS. 2010. Factors affecting
songbird nest survival in Northern mixed-grass prairie. J. Wildl. Manage. 74:257-
59


264.
Kerschner EL, Bollinger EK. 1996. Reproductive success of grassland birds at east-central Illinois airports. Am. Midi. Nat. 136:358-366.
Keyel AC, Strong AM, Perlut NG, Reed JM, Reed AJM. 2013. Evaluating the roles of visual openness and edge effects on nest-site selection and reproductive success in grassland birds. Auk 130:161-170.
Klazura GE, Imy D a. 1993. A Description of the Initial Set of Analysis Products Available from the NEXRAD WSR-88D System. Bull. Am. Meteorol. Soc. 74:1293-1311.
Klicka J, Zink RM, Winker K. 2003. Longspurs and snow buntings : phylogeny and biogeography of a high-latitude clade ( Calcarius ). Sci. Direct 26:165-175.
Klug PE, Jackrel SL, With KA. 2010. Linking snake habitat use to nest predation risk in grassland birds: the dangers of shrub cover. Oecologia 162:803-813.
Knopf FL. 1996. Prairie Legacies—Birds. In: Samson FB, Knopf FL, editors. Prairie Conservation. Washington, D.C.; Covelo, California, p. 135-148.
Kunz M, Kugel PIS. 2015. Detection of hail signatures from single-polarization C-band radar reflectivity. Atmos. Res. 153:565-577.
Laake J, Rexstad E. 2014. RMark - an alternative approach to building linear models in
MARK. In: Cooch E, White G, editors. Program MARK: a gentle introduction. 13th ed. p. C-3-C-113.
Labisky RF. 1957. Relation of hay harvesting to duck nesting under a refuge-permittee system. J. Wildl. Manage. 21:194-200.
Lauenroth WK, Sala OE. 1992. Long-term forage production of North American shortgrass steppe. Ecol. Appl. 2:397-403.
60


Lee CA, Lauenroth WK. 1994. Spatial Distributions of Grass and Shrub Root Systems in the Shortgrass Steppe. Am. Midi. Nat. 132:117-123.
Lehman CP, Rumble MA, Flake LD, Thompson DJ. 2008. Merriam’s Turkey Nest Survival and Factors Affecting Nest Predation by Mammals. J. Wildl. Manage. 72:1765-1774.
Lloyd JD, Martin TE. 2005. Reproductive success of Chestnut-Collared Longspurs in native and exotic grassland. Condor 107:363-374.
Ludlow S, Brigham R, Davis S. 2014. Nesting ecology of grassland songbirds: effects of predation, parasitism, and weather. Wilson J. ... 126:686-699.
Lusk JS, Koper N. 2013. Grazing and songbird nest survival in Southwestern Saskatchewan. Rangel. Ecol. Manag. 66:401-409.
MacDonald MA, Bolton M. 2008. Predation of Lapwing Vanellus vanellus nests on lowland wet grassland in England and Wales: Effects of nest density, habitat and predator abundance. J. Omithol. 149:555-563.
Mahoney K, Alexander M a., Thompson G, Barsugli JJ, Scott JD. 2012. Changes in hail and flood risk in high-resolution simulations over Colorado’s mountains. Nat. Clim. Chang. 2:125-131.
Martin P a., Forsyth DJ. 2003. Occurrence and productivity of songbirds in prairie farmland under conventional versus minimum tillage regimes. Agile. Ecosyst. Environ. 96:107-117.
Martin PA, Johnson DL, Forsyth DJ. 1998. Indirect effects of the pyrethroid insecticide deltamethrin on. Ecotoxicology 7:89-97.
Mason N a., Title PO, Cicero C, Burns KJ, Bowie RCK. 2014. Genetic variation among western populations of the Homed Lark (Eremophila alpestris) indicates recent
61


colonization of the Channel Islands off southern California, mainland-bound dispersal, and postglacial range shifts. Auk 131:162-174.
Mayfield H. 1961. Nesting success calculated from exposure. Wilson Bull. 73:255-261.
Mayfield HF. 1975. Suggestions for Calculating Nest Success. Wilson Bull. 87:456-466.
McKechnie AE, Hockey PAR, Wolf BO. 2012. Feeling the heat: Australian landbirds and climate change. Emu 112:i—vii.
McKechnie AE, Wolf BO. 2010. Climate change increases the likelihood of catastrophic avian mortality events during extreme heat waves. Biol. Lett. 6:253-6.
Mclaughlin ME, Janousek WM, Mccarty JP, Wolfenbarger LL. 2014. Effects of urbanization on site occupancy and density of grassland birds in tallgrass prairie fragments. J. F. Ornithol. 85:258-273.
Mickey FW. 1943. Breeding Habits of McCown’s Longspur. Auk 60:181-209.
Milchunas DG, Lauenroth WK, Chapman PL, Mohammad KK. 1989. Effects of Grazing, Topography, and Precipitation on the Structure of a Semiarid Grassland Author(s): Daniel G. Milchunas, William K. Lauenroth, Philip L. Chapman and Mohammad K. Kazempour Source: Vegetatio 80:11-23.
Miller HJ. 2004. Tobler’s first law and spatial analysis. Ann. Assoc. Am. Geogr. 94:284-289.
Mineau P, Whiteside M. 2013. Pesticide acute toxicity is a better correlate of U.S. grassland bird declines than agricultural intensification. PLoS One 8.
Morgan GMJ, Towery NG. 1975. Small-scale variability of hail and its significance for hail prevention experiments. J. Appl. Meteorol. Climatol. 14:763-770.
Murray L. 2015. Success and Predation of Bird Nests in Grasslands at Valley Forge National
62


Historical Park. Northeast. Nat. 22:10-19.
Newton I. 2007. Weather-related mass-mortality events in migrants. Ibis (Lond. 1859). 149:453-467.
Nice MM. 1957. Nesting success in altricial birds. Auk 74:305-321.
Paquin D, de Elia R, Frigon A. 2014. Change in North American Atmospheric Conditions Associated with Deep Convection and Severe Weather using CRCM4 Climate Projections. Atmosphere-Ocean 52:175-190.
Perkins DG, PerlutNG, Strong AM. 2013. Minor fitness benefits for edge avoidance in nesting grassland birds in the Northeastern United States. Auk 130:512-519.
Perlut NG, Strong AM, Alexander TJ. 2011. A model for integrating wildlife science and
agri-environmental policy in the conservation of declining species. J. Wildl. Manage. 75:1657-1663.
Peterson A, Young H. 1950. A nesting study of the Bronzed Grackle. Auk 67:466-476.
Pietz PJ, Granfors DA. 2000. Identifying predators and fates of grassland passerine nests using miniature video cameras. J. Wildl. Manage. 64:71-87.
Pleszczynska WK. 1978. Microgeographic Prediction of Polygyny in the Lark Bunting. Science (80-. ). 201:935-937.
Polak M, Starmer WT. 2005. Environmental origins of sexually selected variation and a
critique of the fluctuating asymmetry - sexual selection hypothesis. Evolution (N. Y). 59:577-585.
Pons P, Lambert B, Rigolot E, Prodon R. 2003. The effects of grassland management using fire on habitat occupancy and conservation of birds in a mosaic landscape. Biodivers. Conserv. 12:1843-1860.
63


Pulliam HR. 1988. Sources Sinks and Population Regulation. Am. Nat. 132:652-661.
Robel RJ, Briggs JN, Dayton AD, Hulbert LC. 1970. Relationships between visual
obstruction measurements and weight of grassland vegetation. J. Range Manag. 23:295-297.
Rosenberg NJ. 1987. Climate of the Great Plains region of the United States. Gt. Plains Q. Paper 344:22-32.
Rotella JJ, Dinsmore SJ, Shaffer TL. 2004. Modeling nest-survival data: A comparison of recently developed methods that can be implemented in MARK and SAS. Anim. Biodivers. Conserv. 27:187-205.
Rotenberry JT, Wiens J a. 1980. Habitat structure, patchiness, and avian communities in
North American steppe vegetation - a multivariate analysis. Ecology 61:1228-1250. Ryzhanovsky VN. 2015. Comparative ecology of horned lark Eremophila alpestris flava Gm. and snow bunting Plectrophenax nivalis L. in subarctic and arctic zones. Contemp. Probl. Ecol. 8:309-316.
Sala OE, Lauenroth WK, Parton WJ. 1992. Long-term soil water dynamics in the shortgrass steppe. Ecology 73:1175-1181.
Sanderson EW, Redford KH, Aune K, Derr IN, Dobrott S, Fearn E, Fleener C, Forrest S, Gerlach C, Cormack C. 2008. The Ecological Future of the North American Bison : Conceiving Long-Term , Large-Scale Conservation of Wildlife Authors.
Saunders DA, Mawson P, Dawson R. 2011. The impact of two extreme weather events and other causes of death on Carnaby’s Black Cockatoo : a promise of things to come for a threatened species? Pacific Conserv. Biol. 17:141-148.
Schiesser HH. 1990. Hailfall: the relationship between radar measurements and crop
64


damage. Atmos. Res. 25:559-582.
Sexson MG, Farley GH. 2012. Snowy plover nest survival in Kansas and effective
management to counter negative effects of precipitation. J. Wildl. Manage. 76:1587-1596.
Sheldon RD, Chaney K, Tyler GA. 2007. Factors affecting nest survival of Northern Lapwings Vanellus vanellus in arable farmland: an agri-environment scheme prescription can enhance nest survival. Bird Study 3657:168-175.
Skagen S, Yackel Adams AA. 2010. Are there optimal densities for prairie birds? Condor 112:8-14.
Skagen SK, Yackel Adams AA. 2012. Weather effects on avian breeding performance and implications of climate change. Ecol. Appl. 22:1131-45.
Skagen SK, Yackel Adams AA, Adams RD. 2005. Nest survival relative to patch size in a highly fragmented shortgrass prairie landscape. Wilson Bull. 117:23-34.
Slatyer RA, Hirst M, Sexton JP. 2013. REVIEW AND Niche breadth predicts geographical range size : a general ecological pattern. Ecol. Lett. 16:1104-1114.
Stanley TR. 2010. Land use and small mammal predation effects on shortgrass prairie birds. J. Wildl. Manage. 74:1825-1833.
Stauffer GE, Diefenbach DR, Marshall MR, Brauning DW. 2011. Nest success of grassland sparrows on reclaimed surface mines. J. Wildl. Manage. 75:548-557.
Steel PE, Dalke PD, Bizeau EG. 1957. Canada Goose production at Gray’s Lake, Idaho, 1949-1951. J. Wildl. Manage. 21:38-41.
Swain SD. 1991. Metabolism and energy reserves of brood-rearing Homed Larks
(Eremophila alpestris). Comp. Biochem. Physiol. — Part A Physiol. 99:69-73.
65


Tobler WR. 1970. A computer movie simulating urban growth in the Detroit region. Econ. Geogr.:234-240.
Trapp RJ, Diffenbaugh NS, Brooks HE, Baldwin ME, Robinson ED, Pal JS. 2007. Changes in severe thunderstorm environment frequency during the 21st century caused by anthropogenically enhanced global radiative forcing. Proc. Natl. Acad. Sci. 104 : 19719-19723.
VanBeek KR, Brawn JD, Ward MP. 2014. Does no-till soybean farming provide any benefits for birds? Agric. Ecosyst. Environ. 185:59-64.
Vickery PD, Hunter ML, Wells J V. 1992. Evidence of Incidental Nest Predation and Its Effects on Nests of Threatened Grassland Birds. Oikos 63:281-288.
Vos SM, Ribic CA. 2013. Nesting success of grassland birds in oak barrens and dry prairies in West Central Wisconsin. Northeast. Nat. 20:131-142.
Warren KA, Anderson JT. 2005. Grassland songbird nest-site selection and response to mowing in West Virginia. Wildl. Soc. Bull. 33:285-292.
Wauer RH, Wunderle JMJ. 1992. The effect of Hurracane Hugo on bird populations on St. Croix, U.S. Virgin Islands. Wilson Bull. 104:656-673.
Weidman T, Litvaitis JA. 2011. Are small habitat patches useful for grassland bird conservation? 2011 18:207-216.
Westerskov K. 1950. Methods for determining the age of game bird eggs. J. Wildl. Manage. 14:56-67.
White GC, Burnham KP. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46:S120-S139.
Wiley JW, Wunderle JM. 1993. The effects of hurricanes on birds, with special reference to
66


Caribbean islands. Bird Conserv. Int. 3:319-349.
Winter M, Johnson DH, Shaffer JA. 2005. Variability in vegetation effects on density and nesting success of grassland birds. J. Wildl. Manage. 69:185-197.
Winter M, Johnson DH, Shaffer JA, Svedarsky WD. 2004. Nesting biology of three
grassland passerines in the Northern tallgrass prairie. Wilson Bull. 116:211-223.
Winter M, Johnson DH, Shaffer JA, Therese M, Svedarsky WD. 2006. Patch size and
landscape effects on density and nesting success of grassland birds. J. Wildl. Manage. 70:158-172.
With K. 1994. The Hazards of Nesting near Shrubs for a Grassland Bird, the McCown’s Longspur. Condor 96:1009-1019.
With KA, King AW, Jensen WE. 2008. Remaining large grasslands may not be sufficient to prevent grassland bird declines. Biol. Conserv. 141:3152-3167.
With KA, Webb DR. 1993. Microclimate of ground nests: the relative importance of
radiative cover and wind breaks for three grassland species. Condor 95:401-413.
Wylie BK, Meyer DJ, Tieszen LL, Mannel S. 2002. Satellite mapping of surface biophysical parameters at the biome scale over the North American grasslands a case study. Remote Sens. Environ. 79:266-278.
Yackel Adams AA, Skagen SK, Savidge JA. 2006. Modeling Post-Fledging Survival of Lark Buntings in Response to Ecological and Biological Factors Published by : Ecological Society of America content in a trusted digital archive . We use information technology and tools to increase productivity and faci. Ecology 87:178-188.
67


Full Text

PAGE 1

i IDENTIFYING AND QUANTIFYING KEY DRIVERS OF NEST SURVIVAL IN SHORTGRASS STEPPE BIRDS by AMBER ROSE CARVER BS/BA , The Evergreen State College , 2008 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 Integrative Biology 2016

PAGE 2

ii Amber Rose Carver has been approved for the Department of Integrative Biology by Michael Wunder, Advisor Diana Tomback, Advisor David Augustine April 21 st , 2016

PAGE 3

iii Carver, Amber (MS, Biology) Identifying and Quantifying Key Drivers of Nest Survival in Shortgrass Steppe Birds Thesis directed by Professor Michael Wun der and Professor Diana Tomback ABSTRACT level research on breeding birds at the Central Plains Experimental Range in northern Colorado. This research was carried out through the University of Colorado Denver, in cooperation with the U.S.D.A. Agricultural Research Service and the Rocky Mountain Bird Observatory (Now: Bird Conservancy of the Rockies), under a grant from the Nebraska Game and Parks Commission. The intent of the grant was to obtain data that would inform management of habitat for birds that breed on the North American shortgrass steppe. An emphasis was placed on increasing our understanding of best management practices for species of conservation concern. My project was composed of two units: a nest survival analysis pertaining to the three dominant ground nesting species at the site, and an assessment of remotely sensed data for explaining mortality in several species affected by a severe hail storm. My findings inform management by underscoring the importance of vegetation as a de fining characteristic of breeding niches in shortgrass steppe birds. My findings suggest the importance of nest site vegetation to nest survival in Lark Bunting ( Calamospiza melanocorys ). I p resents preliminary data on two other species with sample sizes t oo small for comprehensive analysis: Horned Lark ( Eremophila alpestris ) and Rhynchophanes mccownii ). The latter is a species of conservation concern in Colorado and Nebraska. My research improves our understanding of the influences to po pulation declines in locally threatened species. Research behind my second

PAGE 4

iv chapter informs monitoring by establishing the strong connection between remotely sensed weather data and weather induced bird mortality. Thus, the second chapter of my thesis satis fies the mandate to identify tools that can be used to inform and improve management. The form and content of this abstract are approved. I recommend its publication. Approved: Michael Wunder and Diana Tomback

PAGE 5

v ACKNOWLEDGEMENTS I thank the Colorado Field Ornithologists and the Denver Field Ornithologists for funding both years of my project. I thank Dr. Susan Skagen from the United States Geologic Survey for her material and intellectual support of the project. I thank my field assistants Syed Asif, Moriah Bell, Aaron Yappert, Arielle McDermott Amos and David DeSimone for their hard work and dedication.

PAGE 6

vi TABLE OF CONTENTS CHAPTER I RELATIVE IMPORTANCE OF VEGETATION AND WEATHER TO THE NEST SURVIVAL OF GROUND NESTING SHORTGRASS STEPPE PASSERINES ................................ .................... 1 Abstract ................................ ................................ ................................ ................................ .............. 1 Introduction ................................ ................................ ................................ ................................ ........ 2 Species Niche and Range Models ................................ ................................ ................................ .. 3 Importance of Nest Survival ................................ ................................ ................................ ........... 5 Causes of Grassland Bird Nest Failure ................................ ................................ ........................... 8 Mediators of Nest Depredation ................................ ................................ ................................ ...... 9 Nest site Niche and Survival ................................ ................................ ................................ ........ 11 Case Example: Birds of the Shortgrass Steppe ................................ ................................ ............ 12 Objectives ................................ ................................ ................................ ................................ ..... 18 Hypotheses ................................ ................................ ................................ ................................ ... 18 Methods ................................ ................................ ................................ ................................ ............ 19 Study Area ................................ ................................ ................................ ................................ .... 19 Nest Location and Monitoring ................................ ................................ ................................ ...... 20 Nest Site Vegetation and Weather ................................ ................................ ................................ 21 Basic Analyses ................................ ................................ ................................ ............................. 22 Objective A: N iche Delineation ................................ ................................ ................................ ... 22 Objective B: Drivers of Nest Survival ................................ ................................ .......................... 23 Results ................................ ................................ ................................ ................................ .............. 23 Nest Site Vegetation and Weather ................................ ................................ ................................ 25 Objective A: Niche Delineation ................................ ................................ ................................ .. 27

PAGE 7

vii Objective B: Drivers of Nest Survival ................................ ................................ .......................... 29 Discussion ................................ ................................ ................................ ................................ ........ 33 II W EATHER RADAR DATA CORRELATE TO HAIL INDUCED MORTALITY IN GRASSLAND BIRDS ................................ ................................ ................................ ......................... 35 Abstract ................................ ................................ ................................ ................................ ............ 35 Introduction ................................ ................................ ................................ ................................ ...... 36 Methods ................................ ................................ ................................ ................................ ............ 40 Study Area ................................ ................................ ................................ ................................ .... 40 Nest Site Meas urements ................................ ................................ ................................ ............... 41 Data Analysis ................................ ................................ ................................ ............................... 43 Results ................................ ................................ ................................ ................................ .............. 45 Discussion ................................ ................................ ................................ ................................ ........ 48 LITERATURE CITED ................................ ................................ ................................ ..................... 53

PAGE 8

1 CHAPTER I RELATIVE IMPORTANCE OF VEGETATION AND WEATHER TO THE NEST SURVIVAL OF GROUND NESTING SHORTGRASS STEPPE PASSERINES Abstract Grassland birds avoid competitive exclusion by occupying niches that differ in space use, diet, and activity budget. Here, I compare two models for grassland bird niche structure and describe the biological relevance of these models. This includes a mechan istic model and an empirical model . The mechanistic model suggests that vertical space divergence is negligible in grassland birds. By contrast, the empirical model suggests that divergence in space in both dimensions (horizon tal and vertical) is important. Nest site selection i s an example of a niche parameter that has both horizontal and vertical components. The way that we study and quan tify the importance of nest site vegetation attributes such density and composition has changed, and enhanced computing power makes it possible to fit complex models to a probability based metric of reproductive outcome . Nest failure is primarily caused by predation, but this is rarely measured directly. Instead, features such as vegetation and weather that media te predation are more frequently measured. On the shortgrass steppe, three ground nesting passerines overlap in habitat preference and may dominate the local breeding bird community: Lark Bunting (LARB) , Horned Lark (HOLA) (MCLO) . Al l three are open cup ground nesting passerines, but the birds differ in several ways. Based on a literature review, I conclude that both LARB and MCLO are nomadic , while HOLA is more site faithful . Previous research suggests that HOLA has a broader niche t han LARB or MCLO . In 2014, I located 207 nests where vegetation was later measured: 167 LARB , 21 HOLA , and 19 MCLO . I found that cover by tall vegetation (midgrass and forbs)

PAGE 9

2 was significantly higher for LARB nests than for the other two species, and only LARB nests were associated with shrubs. Contrary to expectation, LARB appeared to occupy a broader niche than the other two species, with regard to nest site vegetation. I used 253 nests that did not fail due to the direct impact of weather (200 LARB , 28 HOLA , and 25 MCLO) to fit time and weather based models, looking at both immediate and longer term impacts of precipitation and temperature. I concluded that survival for all species increased with successive nest age, decreased with time, and was negati vely but inconsistently influenced by daily precipitation. For a subset of 152 nests where vegetation had been measured (119 LARB, 17 HOLA, and 19 MCLO), I fit models based on nest site vegetation height and composition . In LARB, crown cover by dead vegeta tion and shrubs and cacti , as well as basal cover by bare ground, had similar moderate support from the data, and support for daily precipit ation was comparatively negligible. In HOLA and MCLO, there was no clear support for any vegetation variable, which may be a reflection of small sample sizes. Additional research on these species will improve our understanding of the link between niche and the cost associated with phenotypic variability. Introduction Grassland bird communities have relatively low specie s richness but high representation by a few grassland dependent species (Rotenberry and Wiens 1980; Knopf 1996) . Understanding the features that underpin the composition of these communit ies is challenging, because these communities tend to be highly dynamic as a result of environmental stochasticity driven by drought, wildfire, and grazing. Species n iche and range delineation have been modeled in various ways (Cody 1968; Rotenberry and Wiens 1980; Knopf 1996) . Here, I compare two of these models and highlight birthrate as a demographic parameter of interest . I offer an

PAGE 10

3 alternative model explaining t he relation ship between niche breadth, parent bird phenotype, and nest survival, and I assess the value of this model using two seasons of data on the nest survival of shortgrass steppe passerines. Species Niche and Range Models Both mechanistic and empirical models have been used to explain niche delineation and species distribution in grassland birds. Cody ( 1968) proposed a mechanistic model suggesting that birds avoid competitive exclusion through horizontal and v ertical divergence in habitat use, diet specialization, and differential activity budget (Fig . 1). He contended that divergence in the use of vertical habitat is negligible in grassland birds due the near lack of vertical structure, allowing us to focus on the other three dimensions: horizontal space, diet, and activity . This model is useful in that it allows us to categorize likely modes of niche pecies avoid competitive exclusion, with arrows representing direction of selective force.

PAGE 11

4 divergence , but it is based more on deductive reasoning than on emp irical observation and overlooks many of the complexities associated with dynamic grassland bird communities . In contrast, Knopf ( 1996) presented an empirical model for the distribution of birds on the Great Plains (Fig. 2) . This model is informative in that it synthesizes many observations . Unlike the earlier model, it does not make the assumption that vertical structure is unimportant in grassland systems. Rather, it recognizes the existence of ephemeral emergent vegetation whose distribution is driven in part by spatiotemporally variable grazing pressure. Furthermore, it includes niche breadth and overlap , providing a more complex and biologically precise representation of how Great Plains species are distributed . It provides a framework for considering the requirements of bird species coexisting in a heterogeneous landscape. While it is limited to birds treated by th e model, it could easily be ext ended based This empirical model demonstrates the concept that species distributio n responds to vegetation structure, which in turn is a factor of grazing intensity. Thus, grazing induced habitat heterogeneity regulates community composition in Great Plains birds.

PAGE 12

5 on our understanding of other species life histories. Its primary limitations are that it focus es exclusively on the grazing vegetation relationship and does not show how communities respond to change over the course of time . One of the key features not addressed by either model is the cost of individual phenotype. Population genetic variability results in phenotypic heterogeneity , and i ndividual reproductive fitness in a given set of environmental conditions depends in large part on phenotype. This selective pressure means that individuals should generally be found in the environment where they w ill be reproductively successful (Adams 1908; Grinnell 1917) . If optimal habitat is limited, some individuals will occupy suboptimal habitat, al though survival declines with departure from optimal conditions (Fig. 3 , Holt and Keitt 2005) . Niche breadth is regulated in part by the reproductive penalty associated with deviance from mean phenotype (Slatyer et al. 2013) . In the case of nest site selection, the probability of producing at least one offspring from a nesting effort (nest survival) should decrease with departure from mean nesting conditions, which would include features such as vegetative cover, proximity to other nesting pairs, effort timing, and food availability. Thus, quantifying the mean and variance of nesting conditions for e ach species can inform us as to the biology of each species and the modes of niche divergence. M odeling nest survival dynamics is important not only to understanding how populations respond to habitat, but also to understanding the link between population demography and community flux. Importance of Nest Survival Population demography and stability are regulated by birth, death, immigration, and emigration (Pulliam 1988) . Of these four demographic parameters , birth rate is the most

PAGE 13

6 tractable problem in grassland birds for two reasons. First, breeding season adult distribution is predictable based on species niche (Johnson and Grier 1988) . Although individuals may have low site fidelity (Jones et al. 2007) , species occupancy of a given area can be predicted based on habitat attributes ( e.g., Pons et al. 2003 ; Irvin et al. 2013 ; Mclaughlin et al. 2014) . Second, nests can be located based on adult distribution. Once located, nests can be monitored until effort completion, and fate can be ascertained with a fair degree of certainty. The most challenging aspect of this form of research is deriving a biologically relevant response metric for nest site habitat attributes . U ntil the early metric of avian reproductive outcome was apparent nest success ( e.g., Peterson and Young 1950; Glover 1953; Glover 1956; Labisky 1957; Nice 1957; Steel et al. 1957) . Apparent nest success is defined as the number of known nesting efforts in which at least one offspring survived a nd left the nest. Mayfield ( 1961 , 1975) was the first to publicly point out that apparent nest success is a poor metric of population reproductive response . This is because most methods of nest location rely on observing adult behavior, and adults do not remain near the nest site after a nest has failed. Fig. 3. Conceptual model relating species distribution to demographic parameters (Holt and Keitt 2005). If (a) density independent mortality increases across a gradient in habitat quality (x 1 to x 3 ), density dependent birthrate should increase, suggesting that (b) habitat quality (and hence carrying capacity) is lowest in x 1 .

PAGE 14

7 Many nests fail early in the ne sting effort, so most nest samples are skewed toward nests that avoid failure longer. Mayfield advocated for the use of a probability based metric now termed nest survival (Dinsmore et al. 2 002) that quantifies the likelihood of achieving success rather than the proportion of nests known to have succeeded. Although Mayfield ( 1961) was unquestionably influential (Google Scholar citations=1,596), researchers were slow to adopt his methods and terminology during the first 40 years following publication (Fig.4). s of publica both grown dramatically. This indicates an increased research interest likely fueled by enhanced computing power, which allowed us to mov e simplistic constant survival estimates to consider more subtle patterns in survival (Dinsmore et al. 2002; Jehle et al. 2004) . Specifically, the ability to fit multivariate generalized linear models to survival data has been facilitated by automated computation through programs such as MARK (White and Burnham 1999) and scripting languages such as SAS (Rotella et al. 2004) and R (Laake and Rexstad 2 014) . (Fig. 4 ) suggests either the use of outdated terminology or continued use of a metric that has been shown to have little biological relevance. The results of such studies must be interpreted with caution, as they may be poorly reflexive of actual nest survival dynamics .

PAGE 15

8 Causes of Grassland Bird Nest Failure I conducted a and identified a sample of 173 journal articles directly pertaining to causes of grassland bird nest failure . This sample excluded reviews, meta analyses, and research on marsh or coasta l birds. I found that n early all studies in the sample identified depredation as the primary cause of nest failure. I n agricultural grasslands, other potential causes of low nest survival include trampling by livestock , destruction through mowing , and exposure to pesticides . Trampling was found to be important in some studies (Perlut et al. 2011; Beja et al. 2014; Chaiyarat and Eiam Ampai 2014) but not others (Winter et al. 2004; Johnson et al. 2012; Bleho et al. 2014) . and computation. Publication for many biological topics follows a similar pattern, but few topics have shown such a dramatic change as nest surviva Mayfield (1961), to >100 per year over the most recent decade.

PAGE 16

9 Mowing causes low nest survival in some systems (Kerschner and Bollinger 1996; Casas and Viñuela 2010) but not others (Giocomo et al. 2008) . Pesticide effects are understudied. Martin et al. ( 1998) found th at Chestnut collared Longspur was capable of switching diet in response to reduced abundance of a prey item after pesticide appli cation and that nest survival was not strongly impacted. However, the prevalence of pesticide use in grasslands and concurrent idiopathic declines in grassland birds suggest a possible relationship that should be investigated further (Mineau and Whiteside 2013) . Weather is another understudied inf luence on nest survival. Perlut and Strong (2011) found that a high proportion of Savannah Sparrow nests failed due to flooding during heavy rain . Hartman and Oring ( 2009) found that prior winter precipitation explained flood induced nest failure in Long billed Curlew . Severe weather may directly cause low nest survival and adult mortality in bird s (Wauer and Wunderle 1992; Wiley and Wunde rle 1993; Newton 2007; Frederiksen et al. 2008; McKechnie and Wolf 2010; Saunders et al. 2011; McKechnie et al. 2012) . However, quantifying the link between weather and nest survival is challenging because studies cannot be planned around naturally occurring severe weather events . That is because such events are unpredictable . For further discussion on this and a case example, see Chapter 2. Mediators of Nest Depredation Although de predation is the primary cause of nest failure in grassland birds , predator abundance and activity levels are rarely measured directly. Exceptions include Klug et al. ( 2010) , Stanley ( 2010) , and Murray ( 2015) . For a review on nest camera studies, see Degregorio et al. ( 2014) . Grassland nest predators may include mammals (Stanley 2010;

PAGE 17

10 Murray 2015) , snakes (Klug et al. 2010) , and birds (MacDonald and Bolton 2008) , depending on the location and habitat. R esearchers more often choose to measu re habitat features that mediate predation, such as nest site vegetation (Winter et al. 2004; Warren and Anderson 2005; Stauffer et al. 2011; Johnson et al. 2012) , nesting area patch size (Skagen et al. 2005; Winter et al. 2006; Weidman and Litvaitis 2011; Benson et al. 2013) , patch perimeter area ratio (Helzer and Jelinski 1999) , nest proximity to habitat edge (Bollinger and Gavin 2004; Benson et al. 2013; Key el et al. 2013; Perkins et al. 2013) or to woody vegetation (Vos and Ribic 2013) , nest area habitat (Henningsen and Best 2005; Lloyd and Martin 2005; Winter et al. 2006; Cao et al. 2009; Blank et al. 2011; Hatchett et al. 201 3) , or local management regime (Sheldon et al. 2007; Churchwell et al. 2008; With et al. 2008; Fig. 5. Hypothetical niche distribution. Assuming a range of 0 100% in prevale nce of a given feature ( x ), nest site placement may be independent of that feature (gray), more likely to be placed where that feature has low prevalence (blue) or high prevalence (purple). Nests of species with higher niche breadth (generalists) with resp ect to feature x should be more variable in regards to prevalence of x (orange). For variables important to survival, mean prevalence of x should correspond with highest nest survival and highest number of nests. Prevalence of x (%) Probability Density Function

PAGE 18

11 Johnson et al. 2012; VanBeek et al. 2014) . In addition, they may measure daily weather attributes such as rain fall and temperature (Dinsmore et al. 2002; Skagen and Yackel Adams 2012; Ludlow et al. 2014) , or seasonal attributes such as rainfall (Skagen and Yackel Adams 2012) or drought index (Dreitz et al. 2012) , which are thought to influence predator d istribution and activity level. Nesting density may also be important to survival, perhaps due to increased detection by predators in areas with higher nest density (Aguilar et al. 2008; Skagen and Yackel Adams 2010) or enhanced vigilance in a group nesting situation (MacDonald and Bolton 2008) . Relationships between environmental variables and nest survival depend on species life history (Davis et al. 2006; Kerns et al. 2010; Grant and Shaffer 2012; Lusk and Koper 2013 ) , so n est survival analyses are an important tool for quantifying niche attributes. Nest site Niche and Survival As illustrated by Cody ( 1968) , grassland bird species niche is comprised of time budget, diet, and space use . The latter includes nest site vegetation. Knopf ( 1996) highlighted one of these nest site vegetation and here I focus on this aspect as well. N est site vegetation may strongly influence nest survival but has variable importance depending on species life history (Davis 2005; Davis et al. 2006; Kerns et al. 2010; Grant and Shaffer 2012; Lusk and Koper 2013) . Thus, elucidating the relative importance of different nest vegetation attributes provides information on niche position and breadth (Fig. 5) . The range of nest conditions used by each species as well as the magni tude of decline in nest survival with departure from mean conditions are key attributes of niche (Adams 1908; Grinnell 1917) . Grassland ecosystems are highly dynamic, making the exact distribution of conditions unpredictable

PAGE 19

12 from year to year (Winter et al. 2005) . Most grassland breeding birds are migratory, and upon return to the breeding range they may f ollow any of the three settling patterns proposed by Johnson and Grier ( 1988) : homing, opportunistic settling, or flexible settling. There is evidence to suggest that many species exhibit either opportunistic or flexible settling, resulting in perceived nomadism . In t his way, they track the distribution of conditions rather than returning to the same location and suffering the population level reproductive consequences of breeding in subopt imal conditions ( Fretwell 1986; Igl and Johnson 1999; Winter et al. 2005) . This lends support to the hypothesis proposed in Fig. 5 , because nomadism allows species to achieve niche stability despite environmental stochasticity. Case Example: Birds of the Shortgrass Steppe The shortgrass steppe is a region along the western edge of the North American Great Plains where low precipitation and high evaporative potential (Sala et al. 1992) result in dominance by low stature grasses (Mi lchunas et al. 1989) . The ecosystem is defined by d isturbance induced habitat heterogeneity that was historically driven by a combination of grazing by bison and prairie dog s, as well as periodic drought (Derner et al. 2009) . These stochastic factors occur with different periodicities and distributions , generating a complex shifting habitat pattern that is a key regulator of bird community composition (Rotenberry and Wiens 1980; Knopf 1996; Derner et al. 2009) . Birds of the shortgrass steppe occupy niches that show varying overlap (Rotenberry and Wiens 1980; Knopf 1996) . Here, I focus on three species: Lark Bunting ( Calamospiza melanocorys ), Horned Lark ( Eremophila alpestris ), and Rhynchophanes mccownii ). All three belong to the order Passeriformes and are short distance migrants within the Great Plains . Ecologically, t hey are

PAGE 20

13 similar in that they build open cup ground nests, but they differ in several important ways (Table 1) . I selected these species because they breed in similar habitat, and beca use where they overlap in breeding range they may be the dominant ground nesting species. Lark Bunting is in the family Emberizidae and is thought to be high ly nomadic in the breeding season . Their nomadism has not been documented directly through empir ica l research, but it is supported circumstantially by the widely fluctuating local abundance of Lark Bunting (Augustine and Derner 2015) . This fluctuation is almost certainly due to local movements , because magnitude and frequency of change are greater than might be expected through varying rate of birth and death . Lark Bunting nomadism is also indicated by the fact that they are highly sexually dichromatic (Chaine and Lyon 2008) . Increased ornamentation is a sign of stable access to habitat resources (Polak a nd Starmer 2005) , which in the dynamic shortgrass steppe ecosystem strongly suggests that Lark Buntings track resource availability across the landscape, exhibiting opportunistic or flexible settling rather than homing. Male ornamentation is reinforced by polygyny, because females preferentially mate with males on the best territories even when those males have al ready mated with other females (Pleszczynska 1978) . Lark Bunting occupancy is not tied to vegetation attributes (Rotenberry and Wiens 1980) , suggesting that these birds track other environmental features (such as precipitation or food availability) rather than vegetation structure. Although Lark Bunting is classified as a new world sparrow (Carson and Spicer 2003) , it is different fr om many other shortgrass steppe sparrows in that it nests on the ground rather than in a shrub (Pleszczynska 1978) . I t nest s in taller, denser vegetation than , and the nest is relatively well shaded throughout the day (With and Webb 1993; Lusk and Koper 2013) .

PAGE 21

14 Lark Bunting nest survival increases with median grass height (Skagen and Yackel Adams 2010) , but w eather appears to exert a stronger and more consistent effect on Lark Bunting nest survival than nest site vegetation does (Skagen and Yackel A dams 2012) . Similarly, L ark Bunting post fledging survival is negatively impacted by drought (Yackel Adams et al. 2006) . In summary, Lark Bunting is a l ikely nomadic, polygynous species that tends to nest in tall vegetation but whose reproductive success appears tied more to weather and climate than to nest site vegetation. Table 1. Comparison of Focal Species Species Lark Bunting Horned Lark McCown's Longspur Family Emberizidae Alaudidae Calcariidae Breeding Range throughout the Great Plains throughout the northern hemisphere northwestern Great Plains Breeding Habitat mixed grass, shortgrass, shrubland shortgrass shortgrass Breeding Range Settling Opportunistic or flexible ( probable ) Homing ( probable ) Opportunistic or flexible ( probable ) Peak Breeding on Shortgrass Steppe Late May and Mid June Late April, early May Mid May and Mid June Sex Physical Differences Males ornamented, females not Males larger than females Males ornamented, females not Mating System polygyny Social and genetic monogamy ( probable ) Social monogamy, with potential genetic polyandry ( probable ) Parent Roles Both build, incubate, and provision Females tend nests more than males Only females incubate, but both provision young Nest Vegetation In midgrass, forbs, under shrub Sparse, low vegetation Sparse, low vegetation Drivers of Nest Success Seasonal precipitation (+), daily precipitation ( ), daily temperature (+), grass height (+) unknown Proximity to shrubs ( )

PAGE 22

15 Horned Lark is a Holarctic species in the family Alaudidae. The species breeds in a wide range of habitats, including polar deserts and arctic tun dra (Ryzhanovsky 2015) , alpine tundra (Camfield and Martin 2009) , marine islands (Mason et al. 2014) , desert (Dean and Williams 2004) , and cropland (Beason and Franks 1974) . This suggests high tole rance and adaptability, but inter population variance in genes ( e.g., Mason et al. 2014) and phenotypes ( e.g., Camfield and Martin 2009) has not been comprehensively quantified . Horned Lark is categorized as year round resident on the shortgrass steppe, but may migrate short distances which means that we may not be seeing the same individuals throughout the year (With and Webb 1993 ; Manegold and Sudhaus 2004 ) . On the Great Plains, they are strongly associated with low stature vegetation (Rotenberry and Wiens 1980) . Horned Larks are physiologically adapted to temper ature extremes and low water availability, which may serve as an alternate strategy to nomadism (Davies 1982; Swai n 1991) . S ex plumages are not distinct , suggesting a less polygynous social structure (Polak and Starmer 2005) . They are, however, physic ally dimorphic: males are larger than females and invest less e nergy in nestling provisioning (Du et al. 2015) . The extent of social or genetic monogamy in Horned Lark is unknown, but empirical anecdotes indicate complex social behavior (Camfield et al. 2007) . Horned Lark nests tend to be placed in extremely short vegetation with little or no nest cover , but they are frequently placed in the lee of an emergent f eature such as a cactus pad, rock, or pile of dung (Dubois 1935; Beason and Franks 1974) . This feature offers slight protection from wind but almost no shade , which may indicate a partial reliance on solar radiation for incubation (With and Webb 1993) . Habitat patch size does not affect Horned Lark occupancy (Davis 2004) and has a slightly negative but inconsistent effe ct on nest survival (Skagen et al. 2005) . As in Lark Bunting

PAGE 23

16 patch edge does not affect nest survival in this species (Stanley 2010) . Few studies have looked at nest survival dynamics in Horned Lark, so potential relationships can only be inferred from life history. Nests of this species are highly exposed and have not been documented in dense, tall vegetation. Ho wever, Rotenberry and Wiens (1980) estimated higher niche breadth with respect to vegetation for than for Lar k . With respect to climate, this species also occupies a broad niche. In summary, Horned Lark is a likely non nomadic species that nests in short vegetation and whose reprodu ctive success should be tied more to vegetation than weather. (Klicka et al. 2003) . Like Horned Lark, o ccurrence of this species is typically associated with shorter vegetation than would be expected based on climate alone (Rotenberry and Wiens 1980; Henderson and Davis 2014) , signifying a need for grazing disturbance . It should be noted that Bogard and Davis ( 2014) s Longspur occupancy decreased with percent cover by grass and increased with vegetation height . However, the effects were 0.022 and 0.033 respectively), and do not necessarily contradict the general O ccupancy is more strongly tied to grazing by large, generalist herbivores (Mickey 1943; Felske 1971) and prairie dogs (Augustine and Baker 2013) than to fire (Augustine and Derner 2015) . Like Lark Bunting, the species is sexually dichromati c, with males being more st rongly ornamented than females (DuBois 1937) , suggesting genetic polyandry though they appear to be socially monogamous (Mickey 1943, Felske 1971). A fluctuates widely between years (Martin and Forsyth 2003) , indicating that this species may be nomadic as well .

PAGE 24

17 much more common on rangeland ( Mickey 1943 , Felske 1971 ). Breeding territories have greater coverage by low stature vegetation, more cattle dung, more lichen, and more forbs than generally available habitat , but vegetation does not strongly affect apparent nest success (Greer and Anderson 1989) . It is a semi colonial breeder, nesting in denser aggregations than would be predicted through habitat quality alone (Mickey 1943 , Felske 1971 ) . Females are solely responsible for incubation, but both males and females provision young ( DuBois 1937, Mickey 1943). vegetation and typically have little cover, though they are often placed in the lee of an emergent feature that provides no shade (Dubois 1935; DuBois 1937; Mickey 1943; With and Webb 1993) . Apparent nest success is lower near shrubs , probably due to a higher density of egg eating mammals in areas with perennial woody vegetation (With 1994) . In success is influenced by woody vegetation. Based on the above species profiles, I hypothesize that both Lark Bunting and pressu re and the associated vegetation. I hypothesize that Horned Lark is not nomadic and instead relies on physiological adaptations to cope with fluctuations in local breeding conditions. Longspur (Rotenberry and Wiens 1980), an adaptation that compensates for being non nomadic.

PAGE 25

18 Objectives A) Quantify differences in nest site vegetation height and composition in the three dominant ground nesting species at the site: Lark Bunting, Horned L ark, and B) Quantify relative importance of recent weather including temperature and precipitation to nest survival in each species. C) Quantify relative impact of vegetation characteristics on the nest survival of each species . Hypotheses A) Lark Bunting nests will have significantly higher percent cover by midgrasses, forbs, shrubs, and sub sites will vary more in these features B) Increasing daily precipitation will be decrease nest survival for all species . Rationale due to higher humidity and an increase in odor transmission and predator activity after rainfall events (Dinsmore et al. 2002) . Increasing temperature will increase nest survival for all species, because it will lower the degree of activity needed to maintain metabolic stasis and warm the nest (Kendeigh 1939). This effect will be smaller in Horned Lark, which is m ore physiologically adapted to temperature variation. C) Percent cover by tall vegetation (midgrasses, forbs, shrubs, and subshrubs) will be important to nest survival for all species (Fig. 6) . Rationale: i ncrease in cover by tall

PAGE 26

19 vegetation will lead to a de Longspur, but an increase in nest survival for Lark Bunting. Methods Study Area This study was conducted at Central Plains Experimental Range ( CPER ) , in Weld County, Colorad o on the western edge of the Pawnee National Grassland . This is long term grazing research facility currently administered by the U.S.D.A. Agri cultural Research Service (ARS) . The study area consisted of 129.5 ha pastures segregated by barbed wire fences and Figure 6. Hypothetical distribution of nests with respect to vegetation, conceptually based on figure by Knopf (1996). Assuming a range of 0 (purple) nests will be more similar to one another than they are to Lark Bunting (blue) nests. Horned Lark nests will be more variable than nests of the other two species.

PAGE 27

20 grazed under uniform and adaptive grazing regimes with domestic cattle. The climate is semi arid, with most precipitation occurring in the summer months (Sala et al. 1992) . Long term climate oscillations contribute to inter annual variability in temperature and rainfall (Rosenberg 1987) . Area habitat is c lassified as shortgrass steppe. The area was grazed intensively by American bison ( Bison bison ) prior to their extirpati on (Sanderson et al. 2008) . They grazed in dense, mobile herds, generating a shifting mosaic of heterogeneous habitat. Thus, bird species that breed on the shortgrass steppe evolved in the context of shifting disturbance and high habitat variability (Knopf 1996; Derner et al. 2009) . Prairie dogs continue to act as important gr azers in the region surrounding CPER, but they have been almost entirely removed from CPER in order to isolate the effect of the livestock grazing treatment. Current v egetation at CPER is dominated by C4 shortgrasses (primarily Bouteloua gracilis and B. da ctyloides ), with smaller proportions of cacti, midgrasses, and forbs (Lauenroth and Sala 1992) . The woody plant community includes both tall stature shrubs and low stature sub shrubs (Lee a nd Lauenroth 1994) . Nest Location and Monitoring I gathered data in 2014. I located nests through rope dragging ( Labisky 1957) , focusing my search effort on 3.3 ha permanent plots that are also used for long term vegetation and breeding bird surveys. There are four plots in each pasture, and I included eight pastures in my search effort, for a total of 32 plots (1 05.6 ha). I dragged each plot once per week between May and July, for a total of 7 8 repeat visits per pasture. I also found nests opportunistically . Upon locating a nest, I marked the nest location with a wooden stake, placed ~3 m to the north to avoid di sturbing the nest site. I recorded nest GPS coordinates

PAGE 28

21 and recorded nest species and number of eggs and/or young. I estimated egg developm ent using the egg float method (Westerskov 1950) . I estimated chick development based on extent of pins and feathers, as well as behavior (Jongsomjit et al. 2007) . I returned to each nest every 2 4 days to check the sta tus, ceasing once a nesting effort had evidently ended. If a nest became empty prior the earliest possible fledging date, if the nest lining was pulled up, or if there were egg shell fragments the nest was classified as having failed due to predation. On J une 22 nd , a severe hail storm caused the failure of many nests, and these were not included in the analysis for this chapter. For analyses pertaining to the storm sample, see Chapter 2. Nest Site Vegetation and Weather An ARS field crew measured nest site vegetation as soon as possible af ter a nesting effort had ended. Nests where too much time had elapsed and the vegetation had changed were omitted from the sample. Vegetation quantification included taking Visual Obstruction Readings in four cardinal directions (Robel et al. 1970) and measuring distance to the neares t shrub if there was at least one shrub within 30 meters of the nest. We also measured percent composition by major cover type within one meter of the nest using a modified version of the point intercept method (Canfield 1941) . Briefly, this involved placing a metal quadrat over the nest with the ne st at the center, and with a laser beam perpendicular to the ground counting the number of intercepts by cover type along a grid within the frame. In this way, we quantified both crown and basal cover. Cover types included shortgrass, midgrass, forbs, annu al grass, shrubs (height > 15 cm), subshrubs (height < 15 cm), dead vegetation, cactus, dung, and bare ground. We calculated percent cover by each vegetation type by dividing the

PAGE 29

22 number of intercepts of that type by the total number of counts. We collected data on daily maximum and minimum temperature, as well as total precipitation, from an on site weather station. Basic Analyses I assessed whether vegetation features were normally distributed for each species using a Shapiro Wilks 0.05 as being normally distributed. In order to address the potential for collinearity between vegetation and time caused by plant phenology, I tested for correlation between each vegetation feature and time. In order to a ddress the potential for inter taxa relationships such as competition or dependence, I tested for correlation between pairs of vegetation features . For any highly correlated matched pair, I eliminated all the basal cover variable from consideration. In ord er to quantify immediate and delayed or cumulative weather effects, I determined minimum, maximum, and range in temperature, as well as precipitation, for the preceding day. I also calculated average minimum and maximum temperature over the three days prec eding nest effort conclusion. To identify and eliminate collinearity, I looked for correlation between pairs of weather variables. Objective A: Niche Delineation I used a Kruskal Wallis one way analysis of variance by ranks to assess similarity of vegetat ion features that were not normally distributed for all species, while I used a one way ANOVA to compare normally distributed features, treating features with p 0.05 as having at least one species with a significantly different distribution. I used a pos t hoc Kruskal -

PAGE 30

23 Nemenyi Test to identify pairwise differences between species for those features deemed significantly different for at least one species through the Kruskal Wallis or ANOVA tests, and I examined boxplots to identify direction of difference. Objective B: Drivers of Nest Survival For nest survival analyses, I omitted the 70 nests that had failed due to hail, which left a sample of 253 nests for time and weather based analyses, including 200 Lark Bunting nests, 28 Horned Lark Nests, and 25 McC weather on nest survival for each species. To do so, I initially fit a set of time based generalized linear models using R (version 3.2.2) package RMark (Laake and Rexstad 2014) , which uses the R language interface to run program MARK (White and Burnham 1999) . I assessed (Akaike 1974) data support for the model with weight (w), and relative support for the model with the ratio between weights of two models. The time attributes that I included in the fir st set of models were a time varying factor, a constant effect of time, and the age of the nest. I tested the fit of all possible combinations of these variables, as well as an intercept only model. The purpose of this step was to account for variability i ntroduced by unmeasured background variables. For all species, I used the best time based model as a template for constructing more complex weather based models in which I added a single additional variable, using the set of precipitation variables identif ied as independent. For vegetation based analyses, my sample was further reduced to include only those where vegetation had been measured, Results Over the course of nine weeks (May 19 July 20), I located 319 nests belonging to five

PAGE 31

24 Nighthawk . Only one nest each of the latter two species was found, so these species were omitted from analyses, and I focused on the remaining thre e (Table 2) . Table 2. Nest sample for the three most common species. Lark Bunting Horned Lark Longspur Total Nests Located 257 33 27 319 Veg. Measured 167 21 19 207 Data normally distributed? CC* shortgrass No (p<0.001) No (p=0.02) Yes (p=0.36) CC dead veg. No (p<0.001) Yes (p=0.22) Yes (p=0.12) BC** shortgrass No (p<0.001) Yes (p=0.12) Yes (p=0.21) BC dead veg. Yes (p=0.19 ) Yes (0.81 ) Yes (p=0.16) BC bare No (p<0.001) Yes (p=0.27) Yes (p=0.20) Mean high VOR*** No (p<0.001) Yes (p=0.53) Yes (p=0.29) Mean low VOR No (p<0.001 ) Yes (p=0.27) No (p=0.003) Distance to Nearest Shrub No (p<0.001) Yes (p=0.09) Yes (p=0.05) All other veg. characteristics No (p<0.001) No (p<0.001) No (p<0.001) At least one species different from others? CC Midgrass Yes (p<0.001) CC Forbs Yes (p=0.05) BC Midgrass Yes (p<0.001) BC Annual Grass Yes (p=0.002 ) BC Dung Yes (p=0.003) Mean high VOR Yes (p=0.001) Mean low VOR Yes (p=0.001) Distance to Nearest Shrub Yes (p=0.002) All other veg. characteristics No (p<0.001) Pairwise similarity between species HOLA:MCLO HOLA:LARB MCLO:LARB CC Midgrass p=0.74 p=0.01 p<0.01 CC Forbs p=0.85 p=0.31 p=0.09 BC Midgrass p=0.85 p=0.05 p<0.01 BC Annual Grass p=0.88 p=0.09 p<0.02) Mean high VOR p=0.93 p =0.002 p<0.001 Mean low VOR p=0.15 p <0.001 p=0.44 * Crown Cover ** Basal Cover

PAGE 32

25 Nest Site Vegetation and Weather Vegetation was measured at 65% of nests (Table 2) . C rown cover abundance of shortgrass was normally di but not Horned Lark or Lark Bunting . Crown cover abundance of dead vegetation was normally dist ributed for Horned Lark and , but not Lark Bunting. Basal cover abundance of shortgrass was normally cover of dead vegetation was normally distributed for Lark Bunting, Horned Lark, and Longspur . Basal cover of bare ground was normally distributed for Horned Lark and M . Mean high VOR was normally distributed . Mean low VOR was to the nea not Lark Bunting . No other vegetation features were normally distributed. Horned Lark and I found no correlation between time and percent cover for any vegetation cover type . With the exception of dead vegetation, crown cover features were correlated (r>0.50) with basal cover of the same type, but not with any other feature. Crown and basal cover of de ad vegetation were not strongly correlated (r=0.11). M inimum and m aximum daily temperature showed a weak increase over the course of the season (r=0.59 and 0.63 respectively) , maintaining high day to day variability (Fig. 7). I found that minimum daily tem perature was moderately correlated with minimum (r=0.58) and maximum (r=0.63) temperature from the prior day. It was also moderately correlated with average minimum (r=0.59) and maximum (0.60) *** Visual Obstruction Reading

PAGE 33

26 temperature over the past three days. I kept minimum daily temp erature for analysis and removed the moderately correlated variables from consideration. I found that maximum daily temperature was strongly correlated (r=0.75) with daily temperature range, so I removed daily range from consideration. Daily precipitation was not correlated with any other variable. Temperature range from the prior day was moderately correlated with average temperature range over the preceding three days (r=0.63), so I removed the latter from consideration. Rainfall from the prior day was moderately correlated with average precipitation over the preceding three days (r=0.61), so I removed the latter from con sideration. This left me with five non correlated weather variables: daily minimum and Fig. 7. Maximum temperature increase weakly over the course of the season (r 2 =0.38) but remained highly variable from day to day.

PAGE 34

27 maximum temperature , daily precipitation, temperature range and precipitation on the prior day, and average rainfall over the preceding three days. Objective A: Niche D elineation Vegetative f eatures identified as different for at least one species included: c rown cover abundance of midgrass and forbs ; basal cover abundance of midgrass, annual grasses, dung; both mean low and mean high VOR ; an d distance to the nearest shrub . I omitted subshrub and shrub cover from this comparative analysis, given that only Lark Bunting nests were associated with these features. N est sites of Horned Lark and each other than either was to Lark Bunting in the case of several features (Fig. 8) : crown abundance of midgrass and forbs; basal abundance of midgrass and annual grass; and mean high VOR. Interest ingly, basal cover by MCLO HOLA LARB HOLA MCLO LARB HOLA MCLO LARB Crown Cover by Annual Grass Crown Cover by Forbs Crown Cover by Midgrass Fig. 8. Magnitude and direction of difference for vegetation features identified as having unequal distributions. For several features, vegetation was more Horned Lark (HOLA) than between either of those species and Lark Bunting (LARB). For metrics of tall vegetation such as crown cover abundance of midgrass and forbs, cover was higher in Lark Bunting than in the other two species. Basal cover of annual grass showed the opposite trend, reflecting the higher abundance of sixweeks fescue in areas low in heavily grazed areas that are more likely to attract McCown Horned Lark than Lark Bunting.

PAGE 35

28 (p=0.77) but had low similarity between Lark Bunting and Horne d Lark (0.22). Mean low VOR was n either of the other two pairs. Table 3. Fit for time based models . Lark Bunting Model # of Parameters AICc weight Time + Nest Age 3 532.9 --1.00 time + Nest Age 68 588.5 64.7 0.00 Global Model 69 590.7 66.8 0.00 Nest Age 2 653.8 129.9 0.00 Intercept Only 1 633.7 139.8 0.00 Time 2 665.7 141.8 0.00 time 67 718.2 194.3 0.00 time + Time 68 720.4 196.5 0.00 Horned Lark Time + Nest Age 3 56.8 --0.64 Intercept Only 1 59.1 2.39 0.19 Nest Age 2 60.4 3.7 0.10 Time 2 61.2 4.4 0.07 time + Nest Age 52 184.9 128.1 0.00 Global Model 53 190.1 133.4 0.00 time 51 191.9 135.4 0.00 time + Time 52 197.0 140.2 0.00 Time + Nest Age 3 57.3 --0.59 Time 2 58.7 1.5 0.28 Intercept Only 1 61.3 4.0 0.08 Nest Age 2 62.5 5.2 0.04 time + Nest Age 68 182.4 125.2 0.00 Global Model 69 187.8 130.5 0.00 time 67 191.3 134.0 0.00 time + Time 68 196.7 139.4 0.00 Time: a constant survival trend over time time: a factor by which survival varies from day to day Nest Age: days since a nest effort was initiated Global Model: including all possible parameters

PAGE 36

29 Objective B: Drivers of Nest Survival For all species, I found that a combination of time as a constant effect and nest age m ost strongly explained nest survival (Table 3) . T his relationship was high in all species, with little support for other models. According to the best model in each species , n est survival decreased with exposure time ( LARB: 0.2 8 , CI= 0. 33 : 0.2 2 ; HOLA: 0.22, CI= 0.35: 0.19, CI= 0.31: 0.08 ) , but nest survival increased with nest age ( LARB: 25 , CI=0. 20 :0.2 9 ) . For Lark Bunting, a model including time, nest age, and daily precipitation was best supported by the data, followe d distantly by the global model, and supported 2.5 times more than the global model (Table 4) . According to the best fitting model, daily precipitation had a 0.08, CI = 0.11: 0.04). Relationships were less clear for For Horned Lark, I again found that precipitation was a modest predictor of nest fate . This was followed by time and age only, average rain over the past three days, and temper ature range on the preceding day . Precipitation on the last day had a mostly negative but inconsistent impact on nest 0.13, CI= 0.27:0.00), while average rain over the past three days had a slight positive 0.14:0.39). Temperature range on the previous day had a moderately 0.10, CI= 0.28:0.07). Due to the small sample size and strong risk of spurious results from comparing too many mode ls, I selected a single best weather model by comparing existing models against one in which all top variables were included. This new model did not outperform the rain only model. The relative weight in favor of the latter was 1.4 times that of the more c omplex model.

PAGE 37

30 fitting models: average rain over the past three days , rain on the prior day, and rain on the last day. This was followed by the time only model and the global model. Support was low related models. I again compared existing models against one in which all top variables were included, and that new m odel was best supported . Table 4. Fit for weather based models. All Models include Time and Nest Age. Lark Bunting Model # of Parameters AICc weight Rain 4 511.3 --0.70 Global Model 9 513.2 1.85 0.28 Min 4 518.5 7.18 0.19 Rain1 4 523.6 12.32 0.00 No Weather 3 523.9 12.53 0.00 Max 4 525.3 14.02 0.00 Rain3 4 525.5 14.15 0.00 Range1 4 525.5 14.16 0.00 Intercept Only 1 663.7 152.4 0.00 Horned Lark Min 4 56.7 --0.17 No Weather 3 56.8 0.48 0.16 Range1 4 56.8 0.51 0.16 Rain3 4 57.1 0.35 0.14 Rain 4 57.9 1.18 0.09 Max 4 58.6 1.84 0.07 Rain1 4 58.8 1.03 0.06 Intercept Only 1 59.1 2.43 0.05 Rain + Rain3 + Range 6 59.2 2.50 0.05 Global Model 9 59.4 2.64 0.04 Rain + Rain3 + Rain1 6 53.3 --0.46 Rain 4 54.8 1.57 0.21 Rain3 4 56.4 3.15 0.09 No Weather 3 57.3 4.00 0.06 Rain1 4 58.0 4.78 0.04 Max 4 58.3 5.00 0.04 Min 4 58.3 5.01 0.04 Global Model 9 58.4 5.15 0.03 Range 4 59.4 6.14 0.02 Intercept Only 1 61.3 7.99 0.01

PAGE 38

31 Rain: Total precipitation on the final day of the nest effort Rain1: Total precipitation on the penultimate day of the nest effort Rain3: Total Rain over the three days preceding the end of the nesting effort Min: Minimum temperature on the final day of the nesting effort Max: Maximum temperature on the final day of the nesting effort Range: Temperature range (max min) on the final d ay of the nesting effort No Weather: Time and Nest Age, with no additional parameters Global Model: including all possible parameters 0.43, CI= 0.73: 0.12), precipitation on the penultimate day had a us ually positive but inconsistent impact on 0.34:4.39), and average rain over the past three days had a negative 0.33: 0.06). My sample of nests where vegetation had been measured included 152 nests : 119 For Lark Bunting, I compared the fit of vegetation based models with and without rain. Among the models including vegetation, five had competing support from the data: basal cover by bare ground (w=0.12), crown cover by dead vegetation , crown cover by shrubs 0.93 , w=0.08 ), crown cover by cacti 1.30 , w=0.07 AICc >2) , including daily rainfall . Nest survival showed a slight negative but inconsistent decrease with 0.02, CI= 0.04:0.01), crown cover by dead vegetation 0.0 2 , CI= 0.0 4 0.0 2 , CI= 0.0 5 :0.0 1 ). Nest survival showed a For Horned Lark, I again compared the fit of vegetation based m odels with and without rain. Due to the small sample size, I only built models including crown cover by midgrass, forbs, dead vegetation, and annual grasses; mean low and high VOR; and basal cover by dead vegetation and bare ground. In this case, the globa l model received the most support ( indicating no clear superiority among models. This is likely a spurious result and signifies a

PAGE 39

32 need for a larger sample size. based model against the same set of vegetation based models included for Horned Lark. I found that the rain based model received the most support (w=0.50), while a model including crown cover Table 5. Fit for vegetation based models. All models include Time and Nest Age. Lark Bunting Model # of Parameters AICc weight BC Bare Ground 4 324.8 --0.12 CC Dead Vegetation 4 325.1 0.37 0.10 CC Shrub 4 325.7 0.92 0.08 CC Forb 4 326.0 1.20 0.07 CC Cactus 4 326.0 1.26 0.05 Rain + BC Bare Ground 5 326.6 1.82 0.04 Rain + CC Cactus 5 327.0 2.23 0.04 CC Midgrass 4 327.0 2.27 0.03 BC Dead Veg. 4 327.3 2.51 0.03 CC Annual Grass 4 327.3 2.54 0.03 Rain + CC Shrub 5 327.5 2.76 0.03 Mean Hight VOR 4 327.6 2.85 0.03 Rain 4 327.7 2.89 0.03 Rain + CC Forbs 5 327.8 2.94 0.03 CC Subshrubs 4 327.7 2.94 0.03 Rain + CC Cactus 5 327.8 3.01 0.03 BC Dung 4 328.0 3.18 0.02 Mean Low VOR 4 328.0 3.19 0.02 CC Shortgrass 4 328.0 3.19 0.02 Rain + CC Midgrass 5 328.8 3.97 0.02 Rain + BC Dead Veg. 5 329.0 4.20 0.01 Rain + CC Annual 5 329.1 4.31 0.01 Rain + Mean High VOR 5 329.3 4.54 0.01 Rain + CC Shubshrub 5 329.4 4.65 0.01 Rain + Shrub Distance 5 329.5 4.70 0.01 Rain + BC Dung 5 329.5 4.70 0.01 Rain + Mean Low VOR 5 329.7 4.88 0.01 Rain + CC Shortgrass 5 329.7 4.89 0.01 Global Model 14 329.7 11.16 0.00 Intercept Only 1 409.6 84.82 0.00 Horned Lark Global Model 12 37.9 --1.00 CC Annual Grass 4 56.2 18.2 0.00 Mean Low VOR 4 57.3 19.4 0.00 Rain 4 57.9 20.0 0.00 BC Bare Ground 4 58.4 20.5 0.00 CC Midgrass 4 58.6 20.6 0.00 BC Dead Vegetation 4 58.6 20.7 0.00 CC Forbs 4 58.8 20.9 0.00 CC Dead Vegetation 4 58.8 20.9 0.00 Mean High VOR 4 58.8 20.9 0.00 Intercept Only 1 59.1 21.2 0.00

PAGE 40

33 Rain + Rain1 + Rain3 6 53.3 --0.51 CC Dead Vegetation 4 54.8 1.55 0.23 CC Midgrass 4 57.0 3.77 0.08 BC Bare Ground 4 58.0 4.74 0.05 Mean Low VOR 4 59.1 5.83 0.03 CC Annual Grass 4 59.3 6.05 0.02 Table 5. (Continued) Mean High VOR 4 59.4 6.01 0.02 CC Forbs 4 59.4 6.10 0.02 BC Dead Vegetation 4 59.4 6.14 0.02 Intercept Only 1 61.3 0.01 0.01 Global Model 14 67.8 0.00 0.00 BC: Basal Cover CC: Crown Cover VOR: Visual Obstruction Reading Rain: Total precipitation on the final day of the nest effort Rain1: Total precipitation on the penultimate day of the nest effort Rain3: Total Rain over the three days preceding the end of the nesting effort Min: Minimum temperature on the final day of the nesting effort Max: Maximum temperature on the final day of the nesting effort Range: Temperature range (max min) on the final day of the nesting e ffort Global Model: including all possible parameters by dead vegetation received second In the best model, rain on the final day was indicated as having a very strong and unam 0.79 +/ 0), which is undoubtedly a spurious result, calling into question the rest of the impact estimates. Discussion one another than they are to nests of Lark Bunting, supporting the contention that the first two species nest primarily in shortgrass while Lark Bunting nests in taller grass. Contrary to expectation, I found that Horned Lark has lower niche breadth in most re spects than Lark Bunting, suggesting that Lark Bunting is in fact a generalist while Horned Lark is more of a specialist. This indicates that, in addition to be ing a probable nomad, Lark Bunting gambles regarding nest site vegetation, assuming greater risk at an individual level but perhaps

PAGE 41

34 occupy narrow niches with respect to vegetation, indicating a lower assumption of risk at the individual level but a greater risk o f population decline under sudden environmental change.

PAGE 42

35 CHAPTER II WEATHER RADAR DATA CORRELATE TO HAIL INDUCED MORTALITY IN GRASSLAND BIRDS Abstract Hail contributes to avian mortality . Global climate change will increase the frequency of hail events , which may adversely influence avian population trends. However, estimating the or population demographics is challenging. Hail events are difficult to predict, and they often occur in locations where birds ar e n ot under scientific observation . Estimates of bird mortality through remote sensing would be useful for population monitoring, but observations of bird mortality are usually too imprecise to connect directly with spatial information on storm intensity. Here, we demonstrate a strong connection between Doppler weather reflectivity and nest fate (n=204) during an extreme hail storm that intercepted our study area on 22 June 2014. We provide strong evidence that high values of Doppler reflectivity correspond to hail induced mortality in grassland birds, and that it can thus be used to estimate mortality at multiple scales. This event was part of a larger , ongoing study on grassland bird nest survival at the Central Plains Experimental Range (Weld Co., CO). Th e 2 014 hail storm resulted in high but variably distributed mortality among passerines. We attributed the spatial pattern of mortality to a heterogeneous distribution of hail size s , and we hypothesized that by serving as a proxy for hail size Doppler refle ctivity would accurately explain mortality distribution. To test this, we compared the spatial distribution of nest mortality to the spatial distribution of Doppler weather reflectivity at the time of the event . Average reflectivity was five decibels highe r for locations where nests failed than for locations where nests survived; a threshold value of 62.5

PAGE 43

36 dBZ correctly predicted 96% of nest fates. We conclude that Doppler reflectivity can be used to remotely estimate hail induced nest mortality in populatio ns of grassland birds. This will become an increasingly important tool as hail frequency grows under global climate change. Key Words Doppler data; severe weather; hail; grassland birds; nest mortality.. Introduction Severe weather can kill birds (Wauer and Wunderle 1992; Wiley and Wunderle 1993; Newton 2007; Frederiksen et al. 2008; McKechnie and Wolf 2010; Saunders et al. 2011; McKechnie et al. 2012) t he past, severe weather may have had a negligible long term impact on bird populations, because bird species evolved in the context of weather events . However, under global climate change , the frequency and intensity of severe weather are increasing (Fischer and Knutti 2015) ability to compensate for and recuperate from these events. Hail is a common severe weath er phenomenon in North America (Changnon 2008) , and it can cause high bird mortality in areas where individuals are concentrated (Diehl et al. 2014) . A species whose range is limited to a small, hail prone area can experience a significant population decline due to even a single major event (Saunders et al. 2011) . Thus, we can expect for hail to disproportionately affect species that breed in habitat islands, where adults and young are concentrated with in a small area and the surrounding habitat matrix serves as a poor population reservoir (Duelli and Obrist 2003) . The North American shortgrass steppe is an example of a habitat type reduced to noncontiguous islands because of historical land use and habitat heterogeneity. This region is r egularly subject to hail (Fig. 7 ). This habitat type occurs along the semi arid western edge

PAGE 44

37 of the Great Plains, where a combination of low annual precipitation and high evaporative potential (Sala et al. 1992) resul t in a plant community dominated by low sta ture grasses (Milchunas et al. 1989) . Several bird species breed primar ily on the shortgrass steppe, and nests of most are constructed on or near the ground with little vegetative cover (Knopf 1996) , making them vulnerable to severe weather. Based on climate, the potential range of the shortgrass steppe runs from northern New Mexico and Texas to southern Saskatchewan, but habitat across much of this range has been converted to agri culture (Wylie et al. 2002) or altered through the disruption of the native grazing regime (Derner et al. 2009). The Figure 7 . Study Area Context. climatic potential distribution for the shortgrass steppe (green). The Central Plains Experimental Range (blue star ) is located within both zones.

PAGE 45

38 shortgrass steppe is now limited to islands of semi natural habitat such as the Pawnee National Grassland in northern Colorado. The Pawnee National Grassland is located in the North American hail belt, a broad region where hail occurs with higher frequenc y than in other parts of the continent (Allen et al. 2015). Thus, hail is not an abnormal contributor to mortality for the birds that breed there. However, the frequency of damaging hail storms has increased, and many climate models suggest that this trend will continue into the foreseeable future (Trapp et al. 2007; Kapsch et al. 2012; Paquin et al. 2014; Allen et al. 2015) (e.g., Trapp et al. 2007, Kapsch et al. 2012, Paqu in et al. 2014, Allen et al. 2015). A t least one model , however, predicts a decrease in the frequency of damaging hail over Colorado as the surface warms (Mahoney et al. 2012) . This highlights the uncertainty associated with predicting future climate but unders cores the need for an accurate metric to remotely estimate hail impact as climate change progresses. The ability to e stimat e hail induced mortality may mean the difference between effective and null management of grassland bird habitat. This is challenging because storm location and intensity are heterogeneously distributed (Morgan and Towery 1975) and are not easily predicted (Clark et al. 2012) . Thus , studies cannot be designed to evaluate the impacts of naturally occurring extreme weather events; frequently the zone of highest intensity within a weather system may occur where birds are not under scientific observation. Quantifying the impact of hail on bird populations requires the right remote sensing technology that can be applied at a large spatial scale and can capture local variability in storm location and intensity. The metrics produced must distingui sh between rain and hail, as well as between small and large hail, as these different hydrometeor classes may cause very different outcomes for breeding birds. Here, we introduce Doppler weather radar as a tool for remotely

PAGE 46

39 capturing high resolution, large scale information about storms, and we present a case example demonstrating the connection between Doppler reflectivity and hail induced mortality in breeding birds in a shortgrass steppe habitat island. The United States Doppler weather radar network pro vides high coverage data on atmospheric conditions. Stationary active sensors emit pulses of electromagnetic radiation at multiple angles above the ground, and the sensors then measure the intensity of returning radiation (reflectivity) in decibels relativ e to a standard raindrop (dBZ). Reflectivity is positively correlated with the density of precipitation, and it can be used to reliably distinguish between pure rain and rain mixed with hail (Kunz and Kugel 2015) . Reflectivity increases with hail diameter (Hohl et al. 2002; Hazenberg et al. 2011) . Doppler ra dar sensors emit radiation in a fixed beam width measured in degrees. Resolution (number of unique reflectivity values per unit area) has increased with the deployment of the Next Generation Radar (NEXRAD) network, which uses a narrower beam than the first generation radars (Klazura and Im y 1993) . Still, resolution decays with distance from the emitter, and this must be taken into account when estimating hail impact. Here, we demonstrate the application of NEXRAD data for estimating local storm intensity, as well as the use of station di stance for quantifying spatial resolution. This study was based on nest survival data gathered during the first year of a long term nest monitoring project at the Central Plains Experimental Range (CPER), a shortgrass steppe grazing research facility adja cent to the Pawnee National Grassland. We analyzed the distribution of nest mortality and NEXRAD reflectivity after a severe hail storm in 2014 that had resulted in mass mortality of breeding birds. Our objective was to quantify the accuracy of reflectivit y for explaining hail induced nest mortality resulting from this single event. In

PAGE 47

40 so doing, we assessed the value of this tool for remotely estimating the impact of hail on known populations of breeding grassland birds. Methods Study Area The CPER is located in Weld County, Colorado, on the west unit of the Pawnee National Grassland. The climate is semi arid, with most precipitation occurring in the summer months (Sala et al. 1992) . Severe weather including extreme rain and hail is a common feature of summertime weather, but its distribution is heterogeneous and unpredictable (Rosenberg 1987) . Vegetation at the CPER is dominated by low stature grasses, with smaller proportions of cacti, midgrasses, and forbs (Lauenroth and Sala 1992) . The woody plant communit y includes both tall stature shrubs and low stature subshrubs (Lee and Lauenroth 1994), which might be expected to afford protection from hail. Most bird species that breed at the CPER belong to the order Passeriformes. This includes Lark Bunting ( Calamospiza melanocorys ), Horned Lark ( Eremophila alpestris ), Western Meadowlark ( Sturnella neglecta Rhynchophanes mccownii ), Grasshopper Sparrow ( Ammodramus savannarum Spizella breweri ). These are small bodied organisms that should not be able to survive the physical impact of large hail. The dominant nesting mode at the site is a partially recessed cup in the ground with little or no woody vegetation over the nest. Ground nests are often placed beside an emerge nt clump of grass or cactus, which may afford protection from solar radiation, but not from wind or precipitation (With and Webb 1993) . Very few nests are covered by the type of vegetation that would pro tect nests from the impact of hail.

PAGE 48

41 Nest Site Measurements From May July 2014, we located nests by rope dragging and completed high day interval) nest checks, as part of a larger study examining the relationship between nest survival and li vestock grazing regime. We monitored 204 nests that were active just prior to the hail storm, and analyses reported here pertain to that sample. This included 186 Lark Grasshopper checked again the morning after the storm. We measured nest site vegetation as part of the la rger study, but we were not able to measure vegetation at every nest. Due to logistical constraints, we only measured vegetation at 160 of the 204 hail storm nests. This included er Vegetation based analyses were applied to that subset. The storm occurred 22 June 2014, and a cell producing large hail crossed our study area along a northwest to sout heast trajectory (~1800 1900 MST). Spotters reported hail up to 3.18 cm in diameter, and the volume of hail was estimated to be relatively high. Hail size and force were sufficient to break windows on buildings at the study area. When checking the fate o f nests after the storm, we treated any failed nest as having failed due to the storm. In most cases, failed nests contained crushed eggs or dead chicks. At many nest sites, we also found a dead adult on or near the nest. A few of the nests were empty, but we treated these as having failed due to the storm and assumed that the contents had been scavenged during the several hours since the passing of the storm. We classified nests with live chicks, whole eggs,

PAGE 49

42 and/or a live adult on the nest as still active, and we verified the survival of these nests through continued monitoring. To quantify storm intensity in relation to our nest distribution , we obtained National Weather Service NEXRAD Base Reflectivity at the 0.5 elevation angle for 22 June 2014 from NOA A Climate Data Online (https://www.ncdc.noaa.gov/cdo web/). The data were collected at the National Weather Service station in Cheyenne, Wyoming. Distance from the weather station to nests in the study area ranged from 27 to 37 km. Doppler beam width was 0 .925°. In order to quantify reflectivity resolution, we calculated both mean distance between nests and reflectivity grid cell width. To achieve the former, we imported nest coordinates to QGIS (version 2.8.1 Wien) and calculated mean distance between nest s. To achieve the latter, we estimated reflectivity value grid cell width as the length of an arc with an inside angle equal to the beam width and a radius equal to distance from the station. We calculated reflectivity resolution as the number of grid cell s (and hence unique values) per square kilometer. In order to do so, we quantified the number of nests based on mean distance between nests that would fit inside a grid cell. We assumed that reflectivity would increase with hydrometeor size and density and that other atmospheric phenomena known to reflect Doppler radiation would not have readings exceeding those of the largest hail produced by the storm. For each nest site, we determined maximum reflectivity. We hypothesized that terrain and vegetation woul d be important to nest survival. The terrain variables that we focused on were elevation, slope, and aspect, and we expected to find an interaction effect among terrain variables (Table 6 ). Elevation, slope, and aspect would all affect survival, but the effect of each one would depend on values for the others. For example, nests on a northwest facing slope should have had lower survival than nests on

PAGE 50

43 a southeast facing slope or nests in th e lee of a hill, given that the storm was traveling along a northwest to southeast trajectory. We obtained a 10 m resolution Digital Elevation Model (DEM) (http://coloradoview.org/cwis438/websites/ColoradoView/), and we used it to calculate slope and aspec t for our study area . For each nest site, we determined elevation, slope, and aspect. We further hypothesized that nest site vegetation would strongly influence nest survival. We expected woody vegetation to provide protection, and we quantified this as pe rcent cover by shrubs or subshrubs. We assumed that no other vegetation class (i.e. shortgrass, midgrass, annual grass, forbs, cacti) woul d provide protection from hail. Data Analysis We used a three pronged approach to quantifying the relationship betwee n Doppler reflectivity and nest fate. First, we compared distributions of reflectivity values by fate using Kruskal Wallis one way analysis of variance (Kruskal and Wallis 1952). We assessed magnitude of difference between distributions using the H statis tic and associated p value. Second, we quantified the predictive accuracy of sequential reflectivity values at 0.5 dBZ intervals between the minimum and maximum reflectivity levels measured. For each value, we treated nests associated with equal or greater reflectivity as having failed, and we treated nests associated with lower reflectivity as having survived. We calculated accuracy as overlap between expected and observed numbers of nests in each fate category. In order to assess magnitude of difference b etween expected and observed fate outcomes at each reflectivity level, we used a Chi square goodness of fit test.

PAGE 51

44 Our third approach to analyzing the link between reflectivity and fate was the comparison of fit for generalized linear models based on a pri ori hypotheses regarding the relationship between nest fate and nest site characteristics. We fit two separate sets of models: a terrain and reflectivity based set for the full sample of 204 nests; and a terrain , reflectivity , and vegetation based set for the subset of 160 nests where vegetation had been measured. The first set included 11 models as well as an intercept only and a global model (Table 1). The second set included five models: one each for shrub cover and subshrub cover, another two for shrub or subshrub cover with the best model from the first set, and the global and intercept only. We fit models using R (version 3.2.2) package RMark (Laake and Rexstad 2014) , which uses the R language interface to run program MARK (White and Burnham 1999) models using model weight (w). We quantified relative support for each model using the relative weight in favor of that model, compared to the best supported model. We assessed a whether the estimate was negative or positive and the extent to which the CI overlapped zero. Table 6 . Model Hypotheses and Structure. We based model construction on a priori hypotheses regarding the relationship between nest survival and reflectivity (dbz), elevation (ele), slope (slp), and aspect (asp). Hypothesis Structure declines with increasing reflectivity, as large hail is more likely to kill birds. dbz increases with increasing elevation, as low lying areas are prone to flooding. ele is highest for nests facing away from rather than into the storm. asp is highest for high elevation nests facing away from the storm. ele*asp is highest nests on ste ep slopes facing away from the storm. slp*asp is highest for high elevation nests on steep slopes facing away from the storm. ele*slp*asp increases with increasing elevation but decreases with increasing reflectivity. ele+dbz is highest for nests facing away from the storm in low reflectivity areas. asp+dbz is highest for high elevation nests facing away from the storm in low reflectivity areas. ele*asp+dbz is highest for nests on steep slopes facing away from the storm in low reflectivity areas. slp*a sp+dbz is highest for high elevation nests facing away from the storm on steep slopes in low reflectivity areas. ele*slp*asp+ dbz

PAGE 52

45 Results Mortality and reflectivity were heterogeneously distributed acros s the study area (Fig. 8 ). Reflectivity was on average five decibels higher for nests that failed during the storm than for nests that survived ( H =62.18, p <0.001) (Fig. 9 ). Mean distance between nests was 95.9 m. Reflectivity resolution for the site ranged from 2.27 to 1.67 grid cells per square kilometer. Thus, the number of nests that would be assigned the same value based on distance from the station ranged from 4.6 to 6.2. Explanatory accuracy was highe st (96.1%) for a reflectivity of 62.5 dBZ (Fig. 10 ), and the difference between expected and observed 2 =0.08, p =0.22). Among the reflectivity and terrain based models, the model including reflectivity an d elevation was Figure 8 . Distribution of nest fates and Doppler reflectivity across the study area. More nests failed in the southwestern half of the study area than in the northeastern half. Reflectivity was higher in the southwestern half of the study area than in the northeastern half

PAGE 53

46 most supported by the data (w=0.36), followed by reflectivity alo (Table 7 ). Thus, the best model was 1.71 times more supported by the data than the second best model. Other models were not well supported by the data. Based on the best model, the 0.24, CI=[ 0.31: 0.17]. The effect of elevation 0.12, CI=[ 0.03:0.00]. Within the set of models including cover by shrubs and subshrubs, the reflectivity a nd elevation based model was still best supported (w=0.49), followed by a combination of reflectivity, elevation, and cover by shrubs data than the second best model. A ccording to the best model, the effect of reflectivity was 0.24, CI=[ 0.31: 0.17], but the effect of elevation was neutral 0.03:0.00]. Figure 9 . Distribution of reflectivity by fate. Nests that failed were associated higher and more consistent reflectivity than nests that survived. Failed Survived

PAGE 54

47 Table 7 . Results of model fitting. We fit terrain based models first and then used the best model from that set in subsequent vegetati on based models. These were based on reflectivity (dbz), elevation (ele), slope (slp), aspect (asp), crown cover by shrubs (cshr), and crown cover by shubshrubs (csub). Vegetation based models included both models with the best terrain parameters identifie shown. Support for other models was negligible. Model weight Terrain based Models dbz+ele --0.35 dbz 1.11 0.20 dbz+slp*asp 1.54 0.17 dbz+ele*asp 1.86 0.14 Vegetation based Models dbz+ele --0.49 dbz+ele+cshr 1.48 0.23 dbz+ele+csub 1.89 0.19 Figure 10 . Explanatory accuracy of reflectivity thresholds. Where all nests with a value below a given level are predicted to have survived, while all of those with a value at or above that level are predicted to have failed. Predicted numbers of failed and surviving nests were closest to observed for a reflectivi ty value of 62.5 dBZ. Accuracy was 96.1%.

PAGE 55

48 Discussion High Doppler reflectivity values are a strong indicator of hail induced mortality in grassland birds. This conclusion is based on three pieces of evidence: reflectivity distribution, accuracy of the best reflectivity threshold, and weight in favor of a reflectivity based nest survival model. We therefore assert that reflectivity can be used to remotely estimate hail induced mortality in known populations of grassland birds that may not be under direct observation. Reflectivity may be useful for predicting hail induced mortality in birds breed ing in other habitat types, with the caveat that vegetation and terrain may matter more in other habitats. The impact for birds nesting in more complex nest microhabitat will likely be more strongly tied to vegetation and terrain. Nevertheless, reflectivit y should be a strong predictor of hail induced mortality in most open cup nesters. We found that reflectivity was significantly higher for nests that failed than for nests that survived. Thunderstorm reflectivity is strongly correlated with precipitation r ate (Hazenberg et al. 2011), and levels greater than 55 dBZ indicate hail (Schiesser 1 990) . We assumed that all nest failures associated with this event were a result of direct impact by large hail, and we hypothesized that an increase in reflectivity would be linked to an increase in mortality. A weak connection between reflectivity and fate could have been interpreted either as attributing too many failures to hail when they were due to other causes; or poor capacity for reflectivity alone to predict outcome. Because we found a strong relationship between reflectivity and nest outcome, we concluded that our assumption regarding the agency of mortality was correct for most nests and that our hypothesis regarding the link between reflectivity and fate was supported.

PAGE 56

49 We found that a reflectivity value of 65 dBZ was the tipping point for mor tality and significantly explained nest fate. Despite numerous other storms that year involving high volume rainfall and small hail, no other storm caused direct mortality in nests under observation. Previous studies suggest a rainfall mediated relationshi p between predation and nest mortality ( e.g., Lehman et al. 2008) , but evidence for direct mortality via drowning or nest abandonment appears limited to birds nesting on the ground near waterbodies ( e.g., Sexson and Farley 2012) . Reports have established that hail can kill birds (Hall and Harvey 2007; Newton 2007) (Hall and Harvey 2007, Newton 2007, Diehl et al. 2014), but to our knowledge no other study has quantified the lower size limit for hail that is fatal. The accuracy of our mortality threshold (96%) supports our anecdotal observation that hail size matters. Based on these results, we conclude that small hail associated with reflect ivity in the 55 65 dBZ range is not fatal to small passerines, so hail presence alone is insufficient to explain fate, and size must be taken into account. Nest habitat variables did not explain variation in nest fate. While vegetation may mediate the impa ct of solar radiation in shortgrass birds (With and Webb 1993) , and while woody vegetation appears to play an important role in predation risk (With 1994), its contribution to explaining the distributi on of hail induced mortality in this storm was negligible. Likewise, terrain turned out to be unimportant. Although elevation appeared in the top models from both candidate sets, its influence was neutral, indicating that this may be a spurious artifact. W e included habitat variables in the set of candidate survival models in an effort to account for unexplained variation in nest fate. We thought that shrubs might provide protection from hail, that low lying nests might be more prone to flooding and hail ac cumulation, and that nests on leeward steep slopes might be buffered from hail impact. However, none of these variables had a

PAGE 57

50 consistent association with nest fate, and we concluded that they were not important mediators of hail vulnerability in this syste m. The poor fit of terrain based models to this dataset does not necessarily indicate that terrain is not important to nest survival during hail storms. For example, a cluster of nests that survived in a high reflectivity area may have been protected becau se they were in the lee of a tall hill (Figure 4). The effect of such topographic features may be poorly quantified by simple measurements such as elevation, terrain, and aspect. Accounting for the influence of complex terrain may require more comprehensiv e metrics of proximity to rugged terrain. The survival of nests with more complex nest structure or on more rugged terrain might have a stronger dependence on habitat, in which case the contribution of these features to survival should be quantified. We ex pect that accurate reflectivity based mortality estimates can be derived for most other open cup ground nesting birds. We attribute unexplained variability in nest fate to distance based degradation in reflectivity resolution. United States Doppler weath er technology and coverage have improved substantially over the past 60 years (Brown and Lewis 2005) (Klazu ra and Imy 1993, Brown and Lewis 2005). We can now combine reflectivity with other measurements to more accurately distinguish between rain and hail, as well as estimate hail size (Kunz and Kugel 2015) . Fine scale resolution means that we can also readily distinguish the characteristics of different parts of a storm cell (Hardegree et al. 2008) . H owever, the number of unique values obtainable per unit area is still contingent on distance from the transmitter. Hence, the large size of reflectivity grid cells relative to the mean distance between nests de facto resulted in some inaccurate assignments (Tobler 1970) , stating that simi larity between objects increases with spatial proximity, holds true for

PAGE 58

51 many ecological phenomena (Miller 2004) but appears poorly suited t o the distribution of precipitation within a storm. Therefore, space based binning of nests may have resulted in a large difference between estimated and true hail size for some nests. Hail is important because it may a dd to mortality rates induced by other factors, lowering population viability. A single event may decrease annual production by killing both young and adults. The 2014 hail storm was the primary cause of nest failure that year and resulted in lower estimat ed nest survival than in 2015, a year that was free of major storms. All species covered by this study have the capacity to re nest, so nests destroyed by hail may be replaced by surviving adults within the same season. Conceivably, release from competitiv e reproductive restraint may lead surviving adults to lay bigger clutches, stabilizing productivity for the year (Both 1998) . However, even in the absence of additional stor ms, Fig. 11 . A survival anomaly likely related to terrain. This cluster of twelve nests should have all failed based on local reflectivity. Vegetation data were absent for most of these nests. Survival may have been due to cover by woody vegetation. Alternatively, the apparent contradiction may be an artifact of reflectivity resolution. A third possible explanation is that those nest were protected because they were in the lee of a hill with the highest prominence in the study area. Quantifying topographic context is challenging

PAGE 59

52 probability against success is moderate to high due to the frequency of nest predation in grassland ecosystems ( e.g., Vickery et al. 1992; Pietz and Granfors 2000; Murray 2015) . nest successfully. An alternative possibility is that adults killed by hail may be replaced by individuals immigrating from adjacent areas, restoring the local population size. Yet, this is contingent on proximity to suitable habitat with a reservoir population, and in many cases grassland habitat islands are too isolated for this to be expected. The study also demonstrates the strong link that can exist between remotely sensed weather data and directly observed mortality patterns, promoting the value of Doppler data for estimating the impact of hail events on nesting birds a cross a larger area than that typically encompassed by direct monitoring efforts. Doppler data may serve as an important tool in monitoring bird populations. The frequency of large diameter hail storms has increased due to global climate change, and it is projected to continue increasing. Doppler data make it possible to estimate the extent of large diameter hail impact on nesting birds. Combined with known ranges of breeding birds, it should be possible to estimate hail induced mortality at the regional sc ale. Further, based on climate models and the expected increase in hail frequency, adding hail to the population projection should allow us to generate more accurate estimates for future population sizes given no change in other management parameters. Usin g this information, we can identify changes to management aspects that will most benefit bird populations. It is likely to emerge that increasing contiguity among protected areas will be most beneficial to breeding birds, because that will provide a means to compensate for the effects of high frequency hail events.

PAGE 60

53 LITERATURE CITED /stable/4070690 All use 153. Aguilar TM, Dias RI, Oliveira AC, Macedo RH. 2008. Nest site selection by Blue black Grassquits in a Neotropical savanna: Do choices influence nest success? J. F. Orn ithol. 79:24 31. Akaike H. 1974. A new look at the statistical model identification. IEE Trans. Autom. Control 19:716 723. Allen JT, Tippett MK, Sobel AH. 2015. An empirical model relating U.S. monthly hail occurrence to large scale meteorological environm ent. J. Adv. Model. Earth Syst. 7:226 243. Augustine DJ, Baker BW. 2013. Associations of grassland bird communities with black tailed prairie d in the North American Great Plains. Conserv. Biol. 27:324 334. Augustine DJ, Derner JD. 2015. Patch burn grazing management, vegetation heterogeneity, and avian responses in a semi arid grassland. J. Wildl. Manage. 79:927 936. Beason RC, Franks EC. 1974. Breeding behavior of the Horned Lark. Auk 91:65 74. Beja P, Schindler S, Santana J, Porto M, Morgado R, Moreira F, Pita R, Mira A, Reino L. 2014. Predators and livestock reduce bird nest survival in intensive Mediterranean farmland. Eur. J. Wildl. Res. 60:249 258. Benson TJ, Chiavacci SJ, Ward MP. 2013. Patch size and edge proximity are useful predictors of brood pa rasitism but not nest survival of grassland birds Author ( s ):

PAGE 61

54 887. Blank PJ, Dively GP, Gill DE, Rewa CA. 2011. Bird community response to filter strips in Maryland. J. Wildl. Manage. 75:116 125. Bleho BI, Koper N, Machtans CS. 2014. Direct effects of cattle on grassland birds in Canada. Conserv. Biol. 28:724 734. Bogard HJK, Davis SK. 2014. Grassland songbir ds exhibit variable responses to the proximity and density of natural gas wells. J. Wildl. Manage. 78:471 482. Bollinger EK, Gavin TA. 2004. Responses of nesting Bobolinks (Dolichonyx oryzivorus) to habitat edges. Auk 121:767 776. Both C. 1998. Density dep endence of clutch size: habitat heterogeneity or individual adjustment? J. Anim. Ecol. 67:659 666. Brown R a., Lewis JM. 2005. Path to NEXRAD: doppler radar development at the National Severe Storms Laboratory. Bull. Am. Meteorol. Soc. 86:1459 1470. Camfie ld A, Martin K. 2009. The influence of ambient temperature on horned lark incubation behaviour in an alpine environment. Behaviour 146:1615 1633. Camfield AF, Clason AJ, Martin K. 2007. Mother son parental care in Horned Larks. Wilson J. Ornithol. 119:302 304. Canfield RH. 1941. Application of the line interception method in sampling range vegetation. J. For. 39:388 394. Cao J, He CZ, Wells KMS, Millspaugh JJ, Ryan MR, Ryan MR, Cao J, He CZ, Wells KMS, Millspaugh JJ. 2009. Modeling Age and Nest Specific Sur vival Using a Hierarchical Bayesian Approach. Int. Biometric Soc. 65:1052 1062. Carson RJ, Spicer GS. 2003. A phylogenetic analysis of the emberizid sparrows based on

PAGE 62

55 three mitochondrial genes. Mol. Phylogenet. Evol. 29:43 57. Casas F, Viñuela J. 2010. Agr icultural practices or game management: which is the key to improve red legged partridge nesting success in agricultural landscapes? Environ. Conserv. 37:177 186. Chaine AS, Lyon BE. 2008. Adaptive plasticity in female mate choice dampens sexual selection on male ornaments in the lark bunting. Science (80 . ). 319:459 462. Chaiyarat R, Eiam Ampai K. 2014. Effects of grazing on nesting of the oriental pratincole (Glareola maldivarum) in Bueng Boraphet Wetland, Thailand. Zoolog. Sci. 31:77 82. Changnon SA. 20 08. Temporal and spatial distributions of wind storm damages in the United States. Clim. Change 94:473 482. Churchwell RT, Davis C a., Fuhlendorf SD, Engle DM. 2008. Effects of patch burn management on Dickcissel nest success in a tallgrass prairie. J. Wil dl. Manage. 72:1596 1604. Clark AJ, Weiss SJ, Kain JS, Jirak IL, Coniglio M, Melick CJ, Siewert C, Sobash RA, Marsh PT, Dean AR, et al. 2012. An overview of the 2010 Hazardous Weather Testbed experimental forecast program spring experiment. Bull. Am. Meteo rol. Soc. 93:55 74. Cody ML. 1968. On the methods of resource division in grassland bird communities. Am. Nat. 102:107 147. Davies SJJF. 1982. Behavioural adaptations of birds to environments where evaporation is high and water is in short supply. Comp. Bi ochem. Physiol. -Part A Physiol. 71:557 566.

PAGE 63

56 and vegetation structure on bird abundance and occurrence in southern Saskatchewan. Auk 121:1130 1145. Davis SK. 2005. Nest Site Selection Patterns and the Influence of Vegetation on Nest Survival of Mixed Grass Prairie Passerines. Condor 107:605 616. Davis SK, Brigham RM, Shaffer, TL, James PC. 2006. Mixed grass prairie passerines exhibit weak and variable response s to patch size. Auk 123:807 821. Dean WRJ, Williams JB. 2004. Adaptations of birds for life in deserts with particular reference to Larks (Alaudidae). Trans. R. Soc. South Africa 59:79 91. Degregorio BA, Chiavacci SJ, Weatherhead PJ, Willson JD, Benson TJ, Sperry JH. 2014. Snake predation on North American bird nests: Culprits, patterns and future directions. J. Avian Biol. 45:325 333. Derner JD, Lauenroth WK, Stapp P, Augustine DJ. 2009. Livestock as ecosystem engineers for grassland bird habitat in the western Great Plains of North America. Rangel. Ecol. Manag. 62:111 118. Diehl RH, Bates JM, Willard DE, Gnoske TP. 2014. Bird mortality during nocturnal migration over Lake Michigan: a case study. :19 29. Dinsmore SJ, White GC, Knopf FL. 2002. Advanced te chniques for modeling avian nest survival. Ecology 83:3476 3488. Dreitz VJ, Conrey RY, Skagen SK. 2012. Drought and cooler temperatures are associated with higher nest survival in Mountain Plovers. Du B, Liu C J, Bao S J. 2015. Begging form and growth patt ern of nestlings correlate with parental food allocation patterns in the Horned Lark (Eremophila alpestris). Can. J. Zool. 93:273 279.

PAGE 64

57 238. Dubois BAD. 1935. DuBois, A Dawes. Condo r 37:56 72. Duelli P, Obrist MK. 2003. Regional biodiversity in an agricultural landscape: the contribution of seminatural habitat islands. Basic Appl. Ecol. 4:129 138. matad or, saskatchewan. University of Saskatchewan. Fischer EM, Knutti R. 2015. Anthropogenic contribution to global occurrence of heavy precipitation and high temperature extremes. Nat. Clim. Chang. 5:560 564. Frederiksen M, Daunt F, Harris MP, Wanless S. 2008. The demographic impact of extreme events: stochastic weather drives survival and population dynamics in a long lived seabird. J. Anim. Ecol. 77:1020 1029. Giocomo JJ, Moss ED, Buehler DA, Minser WG. 2008. Nesting biology of grassland birds at Fort Campbel l, Kentucky and Tennessee. Wilson Bull. 120:111 119. Glover FA. 1953. Nesting ecology of the Pied Billed Grebe in Northwestern Iowa. Wilson Bull. 65:32 39. Glover FA. 1956. Nesting and production of the Blue Winged Teal (Anas discors Linnaeus ) in Northwes t Iowa. J. Wildl. Manage. 20:28 46. Grant TA, Shaffer TL. 2012. Time specific patterns of nest survival for ducks and passerines breeding in North Dakota. Auk 129:319 328. 91:609 619. Grinnell J. 1917. The Niche Relationships of the California Thrasher. Auk 34:427 433. Hall DW, Harvey TM. 2007. Mortality at a Night Roost of Great Tailed Grackles and

PAGE 65

58 European Starlings during a Spring Hail Storm. Wilson J. Ornithol. 119:309 312. Hardegree SP, Van Vactor SS, Levinson DH, Winstral AH. 2008. Evaluation of NEXRAD Radar Precipitation Products for Natural Resource Applications. Rangel. Ecol. Manag. 61:346 353. Hartman CA, Oring LW. 2009. Reproductive success of song billed curlews (Nume nius americanus) in Northeastern Nevada hay fields. Auk 126:420 430. Hatchett ES, Hale AM, Bennett VJ, Karsten KB. 2013. Wind turbines do not negatively affect nest success in the Dickcissel (Spiza americana). Auk 130:520 528. Hazenberg P, Yu N, Boudevilla in B, Delrieu G, Uijlenhoet R. 2011. Scaling of raindrop size distributions and classification of radar reflectivity rain rate relations in intense Mediterranean precipitation. J. Hydrol. 402:179 192. Helzer CJ, Jelinski DE. 1999. The Relative Importance o f Patch Area and Perimeter Area Ratio to Grassland Breeding Birds. Ecol. Appl. 9:1448 1458. Henderson AE, Davis SK. 2014. Rangeland health assessment: a useful tool for linking range management and grassland bird conservation? Rangel. Ecol. Manag. 67:88 98 . Henningsen JC, Best LB. 2005. Grassland bird use of riparian filter strips in Southeast Iowa. J. Wildl. Manage. 69:198 210. Hohl R, Schiesser HH, Aller D. 2002. Hailfall: The relationship between radar derived hail kinetic energy and hail damage to buildings. Atmos. Res. 63:177 207. 6. fields: precipitation and patterns of populatio n change. Stud. Avian Biol. 19:178 186. Irvin E, Duren KR, Buler JJ, Jones W, Gonzon AT, Williams CK. 2013. A multi scale

PAGE 66

59 occupancy model for the Grasshopper Sparrow in the Mid Atlantic. J. Wildl. Manage. 77:1564 1571. Jehle G, Yackel Adams AA, Savidge JA, Skagen SK. 2004. Nest survival estimation: a review of alternatives to the Mayfield estimator. Condor 106:472. Johnson DH, Grier JW. 1988. Determinants of breeding distribution of ducks. Lincoln, Nebraska, USA Report No.: 242. Johnson TN, Kennedy PL, Ette rson MA. 2012. Nest success and cause specific nest failure of grassland passerines breeding in prairie grazed by livestock. J. Wildl. Manage. 76:1607 1616. Jones SL, Dieni JS, Green MT, Gouse PJ, The S, Journal W, Mar N, Jones SL, Dieni JS, Green MT, et a l. 2007. Annual Return Rates of Breeding Grassland Songbirds http://www.jstor.org/stable/20455951 REFERENCES Linked references are available NG GRASS. Wilson J. Ornithol. 119:89 94. Jongsomjit D, Jones SL, Gardali T, Geupel GR, Gouse PJ. 2007. A guide to nestling development and aging in altricial passerines. Washington, D.C. Biological Technical Publication Report No.: FWS/BTP R6008 2007. Kaps ch ML, Kunz M, Vitolo R, Economou T. 2012. Long term trends of hail related weather types in an ensemble of regional climate models using a Bayesian approach. J. Geophys. Res. Atmos. 117:1 16. Kerns CK, Ryan MR, Murphy RK, Thompson FR, Rubin CS. 2010. Fact ors affecting songbird nest survival in Northern mixed grass prairie. J. Wildl. Manage. 74:257

PAGE 67

60 264. Kerschner EL, Bollinger EK. 1996. Reproductive success of grassland birds at east central Illinois airports. Am. Midl. Nat. 136:358 366. Keyel AC, Strong AM , Perlut NG, Reed JM, Reed AJM. 2013. Evaluating the roles of visual openness and edge effects on nest site selection and reproductive success in grassland birds. Auk 130:161 170. Klazura GE, Imy D a. 1993. A Description of the Initial Set of Analysis Prod ucts Available from the NEXRAD WSR 88D System. Bull. Am. Meteorol. Soc. 74:1293 1311. biogeography of a high latitude clade ( Calcarius ). Sci. Direct 26:165 175. Klug PE, Jackr el SL, With KA. 2010. Linking snake habitat use to nest predation risk in grassland birds: the dangers of shrub cover. Oecologia 162:803 813. Knopf FL. 1996. Prairie Legacies -Birds. In: Samson FB, Knopf FL, editors. Prairie Conservation. Washington, D.C.; Covelo, California. p. 135 148. Kunz M, Kugel PIS. 2015. Detection of hail signatures from single polarization C band radar reflectivity. Atmos. Res. 153:565 577. Laake J, Rexstad E. 2014. RMark an alternative approach to building linear models in MARK. In: Cooch E, White G, editors. Program MARK: a gentle introduction. 13th ed. p. C 3 C 113. Labisky RF. 1957. Relation of hay harvesting to duck nesting under a refuge permittee system. J. Wildl. Manage. 21:194 200. Lauenroth WK, Sala OE. 1992. Long term f orage production of North American shortgrass steppe. Ecol. Appl. 2:397 403.

PAGE 68

61 Lee CA, Lauenroth WK. 1994. Spatial Distributions of Grass and Shrub Root Systems in the Shortgrass Steppe. Am. Midl. Nat. 132:117 123. Lehman CP, Rumble MA, Flake LD, Thompson DJ and Factors Affecting Nest Predation by Mammals. J. Wildl. Manage. 72:1765 1774. Lloyd JD, Martin TE. 2005. Reproductive success of Chestnut Collared Longspurs in native and exotic grassland. Condor 107:363 374. Ludlo w S, Brigham R, Davis S. 2014. Nesting ecology of grassland songbirds: effects of 699. Lusk JS, Koper N. 2013. Grazing and songbird nest survival in Southwestern Saskatchewan. Rangel. Ecol. Manag. 66: 401 409. MacDonald MA, Bolton M. 2008. Predation of Lapwing Vanellus vanellus nests on lowland wet grassland in England and Wales: Effects of nest density, habitat and predator abundance. J. Ornithol. 149:555 563. Mahoney K, Alexander M a., Thompson G, Barsugli JJ, Scott JD. 2012. Changes in hail and flood risk in high Chang. 2:125 131. Martin P a., Forsyth DJ. 2003. Occurrence and productivity of songbirds in prairie farmland under conventiona l versus minimum tillage regimes. Agric. Ecosyst. Environ. 96:107 117. Martin PA, Johnson DL, Forsyth DJ. 1998. Indirect effects of the pyrethroid insecticide deltamethrin on. Ecotoxicology 7:89 97. Mason N a., Title PO, Cicero C, Burns KJ, Bowie RCK. 2014 . Genetic variation among western populations of the Horned Lark (Eremophila alpestris) indicates recent

PAGE 69

62 colonization of the Channel Islands off southern California, mainland bound dispersal, and postglacial range shifts. Auk 131:162 174. Mayfield H. 1961. Nesting success calculated from exposure. Wilson Bull. 73:255 261. Mayfield HF. 1975. Suggestions for Calculating Nest Success. Wilson Bull. 87:456 466. McKechnie AE, Hockey PAR, Wolf BO. 2012. Feeling the heat: Australian landbirds and climate change. Em u 112:i vii. McKechnie AE, Wolf BO. 2010. Climate change increases the likelihood of catastrophic avian mortality events during extreme heat waves. Biol. Lett. 6:253 6. Mclaughlin ME, Janousek WM, Mccarty JP, Wolfenbarger LL. 2014. Effects of urbanization on site occupancy and density of grassland birds in tallgrass prairie fragments. J. F. Ornithol. 85:258 273. 209. Milchunas DG, Lauenroth WK, Chapman PL, Mohammad KK. 1989. Effects of Grazin g, Topography, and Precipitation on the Structure of a Semiarid Grassland Author(s): Daniel G. Milchunas, William K. Lauenroth, Philip L. Chapman and Mohammad K. Kazempour Source: Vegetatio 80:11 23. . Ann. Assoc. Am. Geogr. 94:284 289. Mineau P, Whiteside M. 2013. Pesticide acute toxicity is a better correlate of U.S. grassland bird declines than agricultural intensification. PLoS One 8. Morgan GMJ, Towery NG. 1975. Small scale variability of hail and its significance for hail prevention experiments. J. Appl. Meteorol. Climatol. 14:763 770. Murray L. 2015. Success and Predation of Bird Nests in Grasslands at Valley Forge National

PAGE 70

63 Historical Park. Northeast. Nat. 22:10 19. Newton I. 2007. Weather relate d mass mortality events in migrants. Ibis (Lond. 1859). 149:453 467. Nice MM. 1957. Nesting success in altricial birds. Auk 74:305 321. Paquin D, de Elía R, Frigon A. 2014. Change in North American Atmospheric Conditions Associated with Deep Convection and Severe Weather using CRCM4 Climate Projections. Atmosphere Ocean 52:175 190. Perkins DG, Perlut NG, Strong AM. 2013. Minor fitness benefits for edge avoidance in nesting grassland birds in the Northeastern United States. Auk 130:512 519. Perlut NG, Strong AM, Alexander TJ. 2011. A model for integrating wildlife science and agri environmental policy in the conservation of declining species. J. Wildl. Manage. 75:1657 1663. Peterson A, Young H. 1950. A nesting study of the Bronzed Grackle. Auk 67:466 476. Pie tz PJ, Granfors DA. 2000. Identifying predators and fates of grassland passerine nests using miniature video cameras. J. Wildl. Manage. 64:71 87. Pleszczynska WK. 1978. Microgeographic Prediction of Polygyny in the Lark Bunting. Science (80 . ). 201:935 93 7. Polak M, Starmer WT. 2005. Environmental origins of sexually selected variation and a critique of the fluctuating asymmetry sexual selection hypothesis. Evolution (N. Y). 59:577 585. Pons P, Lambert B, Rigolot E, Prodon R. 2003. The effects of grassla nd management using fire on habitat occupancy and conservation of birds in a mosaic landscape. Biodivers. Conserv. 12:1843 1860.

PAGE 71

64 Pulliam HR. 1988. Sources Sinks and Population Regulation. Am. Nat. 132:652 661. Robel RJ, Briggs JN, Dayton AD, Hulbert LC. 1970. Relationships between visual obstruction measurements and weight of grassland vegetation. J. Range Manag. 23:295 297. Rosenberg NJ. 1987. Climate of the Great Plains region of the United States. Gt. Plains Q. Paper 344:22 32. Rotella JJ, Dinsmore SJ, Shaffer TL. 2004. Modeling nest survival data: A comparison of recently developed methods that can be implemented in MARK and SAS. Anim. Biodivers. Conserv. 27:187 205. Rotenberry JT, Wiens J a. 1980. Habitat structure, patchiness, and avian communities i n North American steppe vegetation a multivariate analysis. Ecology 61:1228 1250. Ryzhanovsky VN. 2015. Comparative ecology of horned lark Eremophila alpestris flava Gm. and snow bunting Plectrophenax nivalis L. in subarctic and arctic zones. Contemp. Pr obl. Ecol. 8:309 316. Sala OE, Lauenroth WK, Parton WJ. 1992. Long term soil water dynamics in the shortgrass steppe. Ecology 73:1175 1181. Sanderson EW, Redford KH, Aune K, Derr JN, Dobrott S, Fearn E, Fleener C, Forrest S, Gerlach C, Cormack C. 2008. The Conceiving Long Term , Large Scale Conservation of Wildlife Authors. Saunders DA, Mawson P, Dawson R. 2011. The impact of two extreme weather events and promise of things to come for a threatened species? Pacific Conserv. Biol. 17:141 148.

PAGE 72

65 damage. Atmos. Res. 25:559 582. Sexson MG, Farley GH. 2012. Snowy plover nest surviva l in Kansas and effective management to counter negative effects of precipitation. J. Wildl. Manage. 76:1587 1596. Sheldon RD, Chaney K, Tyler GA. 2007. Factors affecting nest survival of Northern Lapwings Vanellus vanellus in arable farmland: an agri envi ronment scheme prescription can enhance nest survival. Bird Study 3657:168 175. Skagen S, Yackel Adams AA. 2010. Are there optimal densities for prairie birds? Condor 112:8 14. Skagen SK, Yackel Adams AA. 2012. Weather effects on avian breeding performance and implications of climate change. Ecol. Appl. 22:1131 45. Skagen SK, Yackel Adams AA, Adams RD. 2005. Nest survival relative to patch size in a highly fragmented shortgrass prairie landscape. Wilson Bull. 117:23 34. Slatyer RA, Hirst M, Sexton JP. 2013. REVIEW AND Niche breadth predicts geographical 1114. Stanley TR. 2010. Land use and small mammal predation effects on shortgrass prairie birds. J. Wildl. Manage. 74:1825 1833. Stauffer GE, Die fenbach DR, Marshall MR, Brauning DW. 2011. Nest success of grassland sparrows on reclaimed surface mines. J. Wildl. Manage. 75:548 557. 1949 1951. J. Wildl. Manage. 21:38 41. Swain SD. 1991. Metabolism and energy reserves of brood rearing Horned Larks (Eremophila alpestris). Comp. Biochem. Physiol. -Part A Physiol. 99:69 73.

PAGE 73

66 Tobler WR. 1970. A computer movie simulating urban growth in the Detroit region. Econ. Geogr.:234 240. Trapp RJ, Diffenbaugh NS, Brooks HE, Baldwin ME, Robinson ED, Pal JS. 2007. Changes in severe thunderstorm environment frequency during the 21st century caused by anthropogenically enhanced global radiative forcing. Proc. Natl. Acad. Sci. 104 :19719 19723. VanBeek KR, Brawn JD, Ward MP. 2014. Does no till soybean farming provide any benefits for birds? Agric. Ecosyst. Environ. 185:59 64. Vickery PD, Hunter ML, Wells J V. 1992. Evidence of Incidental Nest Predation and Its Effects on Nests of Threatene d Grassland Birds. Oikos 63:281 288. Vos SM, Ribic CA. 2013. Nesting success of grassland birds in oak barrens and dry prairies in West Central Wisconsin. Northeast. Nat. 20:131 142. Warren KA, Anderson JT. 2005. Grassland songbird nest site selection and response to mowing in West Virginia. Wildl. Soc. Bull. 33:285 292. Wauer RH, Wunderle JMJ. 1992. The effect of Hurracane Hugo on bird populations on St. Croix, U.S. Virgin Islands. Wilson Bull. 104:656 673. Weidman T, Litvaitis JA. 2011. Are small habitat patches useful for grassland bird conservation? 2011 18:207 216. Westerskov K. 1950. Methods for determining the age of game bird eggs. J. Wildl. Manage. 14:56 67. White GC, Burnham KP. 1999. Program MARK: survival estimation from populations of marked ani mals. Bird Study 46:S120 S139. Wiley JW, Wunderle JM. 1993. The effects of hurricanes on birds, with special reference to

PAGE 74

67 Caribbean islands. Bird Conserv. Int. 3:319 349. Winter M, Johnson DH, Shaffer JA. 2005. Variability in vegetation effects on density and nesting success of grassland birds. J. Wildl. Manage. 69:185 197. Winter M, Johnson DH, Shaffer JA, Svedarsky WD. 2004. Nesting biology of three grassland passerines in the Northern tallgrass prairie. Wilson Bull. 116:211 223. Winter M, Johnson DH, Sha ffer JA, Therese M, Svedarsky WD. 2006. Patch size and landscape effects on density and nesting success of grassland birds. J. Wildl. Manage. 70:158 172. Longspur. Condor 9 6:1009 1019. With KA, King AW, Jensen WE. 2008. Remaining large grasslands may not be sufficient to prevent grassland bird declines. Biol. Conserv. 141:3152 3167. With KA, Webb DR. 1993. Microclimate of ground nests: the relative importance of radiative co ver and wind breaks for three grassland species. Condor 95:401 413. Wylie BK, Meyer DJ, Tieszen LL, Mannel S. 2002. Satellite mapping of surface biophysical parameters at the biome scale over the North American grasslands a case study. Remote Sens. Environ . 79:266 278. Yackel Adams AA, Skagen SK, Savidge JA. 2006. Modeling Post Fledging Survival of Lark Society of America content in a trusted digital archive . We use informa tion technology and tools to increase productivity and faci. Ecology 87:178 188.