ABSTRACT 
 
 
 
 
Title of Dissertation: ECOLOGICAL CAUSES AND 
CONSEQUENCES OF NON-BREEDING 
MOVEMENTS IN A DECLINING 
MIGRATORY SONGBIRD, WOOD THRUSH 
(HYLOCICHLA MUSTELINA) 
  
 Calandra Quinn Stanley, Doctor of Philosophy, 
2019 
  
Dissertation directed by: Dr. Michele Dudash, Department of Biology 
Dr. Peter Marra, Smithsonian Migratory Bird 
Center 
 
 
In this dissertation I evaluate the ecological correlates of non-breeding space-use 
strategies and how these drive within and between season movement dynamics in a 
declining migratory songbird, wood thrush (Hylocichla mustelina). In Chapter 1, I 
deployed high-resolution GPS transmitters across 5 breeding populations to quantify 
habitat selection as wood thrush moved across the annual cycle. I found seasonal 
variation in habitat selection at the regional, landscape and local scales which 
suggests the factors driving the evolution of habitat selection preferences vary across 
seasons and environmental conditions.  In Chapter 2, I combined radio telemetry and 
GPS tracking to examine how environmental conditions drove space-use strategies 
during the non-breeding stationary period. I found evidence that both small- and 
large-scale movement dynamics were dependent on moisture levels on tropical non-
  
breeding grounds. At small spatial scales, dry conditions drove low food availability, 
reduced individual body condition and these individuals had larger home ranges. In 
this same chapter I integrated archival GPS tag data to demonstrate that wood thrush 
from across the breeding range engaged in permanent large-scale mid-winter shifts in 
home ranges and, similar to radio-tagged birds, that individuals from wetter, higher 
quality habitats were more likely to use this strategy. I suggest that the facultative 
movements are a condition-dependent strategy allowing wood thrush to find 
alternative habitats as conditions deteriorate across the dry season in their non-
breeding grounds. Finally, to determine how food availability may influence timing 
of spring migration, I performed a food manipulation experiment with captive wood 
thrush in Chapter 3. I found that food availability and body condition modulated the 
intensity, but not onset, of migratory restlessness (Zugunruhe), an index of migratory 
disposition in captive birds. These results suggest that non-breeding food limitations 
could constrain migration preparation in wood thrush. Low food availability also 
advanced the onset of migratory fattening. I suggest that advancing migratory 
fattening when food availability is low may provide a mechanism to flexibly adjust 
migration timing under poor environmental conditions. Together these findings 
suggest that wood thrush exhibit a diversity of behavioural mechanisms to handle 
environmental heterogeneity across the annual cycle.  
 
  
 
 
 
 
 
 
 
 
 
ECOLOGICAL CAUSES AND CONSEQUENCES OF NON-BREEDING 
MOVEMENTS IN A DECLINING MIGRATORY SONGBIRD, WOOD 
THRUSH (HYLOCICHLA MUSTELINA) 
 
 
 
by 
 
 
Calandra Quinn Stanley 
 
 
 
 
 
Dissertation submitted to the Faculty of the Graduate School of the  
University of Maryland, College Park, in partial fulfillment 
of the requirements for the degree of 
Doctor of Philosophy 
2019 
 
 
 
 
 
 
 
 
 
 
Advisory Committee: 
Emeritus Professor Michele R. Dudash, Co-Chair 
Adjunct Professor Peter P. Marra, Co-Chair 
Professor Gerald S. Wilkinson 
Professor Jiuzhou Song 
Adjunct Professor Helen F. James 
Adjunct Professor Thomas B. Ryder 
  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
? Copyright by 
Calandra Quinn Stanley 
2019 
 
 
 
 
 
 
 
 
 
 
 
 
 
  
 
Dedication 
 
To my family, 
 
For all the love  
and unwavering support over the years. 
 
 
ii 
 
 
 
Acknowledgements 
I would like to thank my advisors, Dr. Michele Dudash and Dr. Peter Marra, for 
taking me on as a student and providing me with the support and guidance to 
complete this project. I would also like to thank Thomas B. Ryder for all his help 
facilitating the WOTH project and for his guidance with my research questions. I 
would also like to thank the rest of my dissertation committee Helen James, John 
Song and Jerry Wilkinson for their insights on the planning and execution of this 
research. All the students and post-docs at the Smithsonian Migratory Bird Center 
and CCEG provided great support and friendship throughout my dissertation. And of 
course, thank you to my fellow BEES students, especially my office mates Grace 
Capshaw and Danielle Adams.  
This work would not have been possible without all the wonderful field 
assistants I worked with through winters in Belize and over the summers in Indiana. 
They made this work so much fun, and I could not have done it without their help. I 
would also like to thank everyone at BFREE for their logistical support and 
friendship and especially Liberato Pop, Marlyn Cruz, William Garcia, Solana Cus, 
Maya Choc, Marcelino Pop, Elmer Tzalam, Nelly Cadle, Sipriano Canti, Judy and 
Dan Dourson, Heather Barrett, Dr. Steven Brewer and last but not least Jacob Marlin. 
I especially want to thank my technicians from the wood thrush captive experiment, 
none of us quite realized what we were getting ourselves into, but with your help we 
were able to make it work! I also need to thank my sister Maia Stanley-Iljic for 
coming out and helping me with the wood thrush in Indiana, Minnesota and DC! And 
none of this would have been possible without the love, support and patience of my 
family. This has all been quite the journey. 
iii 
 
 
 
Table of Contents 
 
 
Dedication ..................................................................................................................... ii 
Acknowledgements ...................................................................................................... iii 
Table of Contents ......................................................................................................... iv 
List of Tables ............................................................................................................... vi 
List of Figures ............................................................................................................. vii 
Overview: ...................................................................................................................... 1 
 Year-round habitat selection for a Neotropical migratory songbird using 
high resolution GPS tags ............................................................................................... 7 
Abstract ..................................................................................................................... 7 
Introduction ............................................................................................................... 8 
Methods................................................................................................................... 14 
Study system ....................................................................................................... 14 
Field Methods and GPS telemetry ...................................................................... 14 
Habitat Features: ................................................................................................. 15 
Local and Landscape Habitat Selection .............................................................. 16 
Regional Habitat Selection ................................................................................. 19 
Statistical Analysis .............................................................................................. 20 
Results ..................................................................................................................... 20 
Patterns of Movement ......................................................................................... 20 
Local and Landscape Habitat Selection .............................................................. 21 
Regional Habitat Selection ................................................................................. 22 
Discussion ............................................................................................................... 24 
Seasonal Variation in Habitat Selection ............................................................. 24 
Regional Habitat Selection ................................................................................. 30 
Conservation Implications .................................................................................. 32 
Acknowledgements ................................................................................................. 33 
Tables ...................................................................................................................... 34 
Figures..................................................................................................................... 36 
Figures Legend.................................................................................................... 36 
 : Patterns of moisture and food availability during the non-breeding season 
influence small- and large-scale space use strategies in a migratory songbird ........... 42 
Abstract ................................................................................................................... 42 
Introduction ............................................................................................................. 43 
Methods................................................................................................................... 46 
Study Species ...................................................................................................... 46 
Field Methods: .................................................................................................... 46 
Radio-telemetry................................................................................................... 47 
Body Condition ................................................................................................... 49 
Diet sampling ...................................................................................................... 49 
Patterns of Moisture ............................................................................................ 51 
GPS Tracking ...................................................................................................... 52 
Data Analysis ...................................................................................................... 53 
Results ..................................................................................................................... 56 
iv 
 
 
 
Home-range Size ................................................................................................. 57 
Food availability and diet.................................................................................... 57 
Body Condition ................................................................................................... 58 
Seasonal Site Persistence .................................................................................... 59 
Discussion ............................................................................................................... 60 
Acknowledgements ................................................................................................. 67 
Tables ...................................................................................................................... 68 
Figures..................................................................................................................... 71 
Figure Legend ..................................................................................................... 71 
 : Food limitation and the endogenous control of spring migration: a captive 
experiment with a long-distance migratory bird ......................................................... 78 
Abstract ................................................................................................................... 78 
Introduction ............................................................................................................. 79 
Methods................................................................................................................... 83 
Study System ...................................................................................................... 83 
Bird Capture and Maintenance ........................................................................... 84 
Body Condition ................................................................................................... 84 
Nocturnal Activity .............................................................................................. 85 
Experimental design............................................................................................ 85 
Timing of Migration ........................................................................................... 86 
Magnitude of Migration ...................................................................................... 87 
Statistics .............................................................................................................. 87 
Results ..................................................................................................................... 88 
Body Condition ................................................................................................... 88 
Timing of Migration ........................................................................................... 89 
Magnitude of Migration ...................................................................................... 89 
Discussion ............................................................................................................... 90 
Acknowledgements ................................................................................................. 96 
Tables ...................................................................................................................... 97 
Figures..................................................................................................................... 98 
Figure Legend ..................................................................................................... 98 
Appendices ................................................................................................................ 104 
...................................................................................................... 104 
 : Seasonal Changes in Habitat Features ...................................... 114 
 : Population-specific annual schedules ....................................... 116 
...................................................................................................... 121 
 : Relationship between stable carbon isotopes and satellite wetness
........................................................................................................................... 134 
Bibliography ............................................................................................................. 140 
 
 
 
 
 
 
v 
 
 
 
List of Tables 
 
Table 1.1. .................................................................................................................... 34 
Table 1.2 ..................................................................................................................... 35 
Table 2.1. .................................................................................................................... 68 
Table 2.2. .................................................................................................................... 69 
Table 2.3 ..................................................................................................................... 70 
Table 3.1 ..................................................................................................................... 97 
Table S1.1. ................................................................................................................ 108 
Table S1.2 ................................................................................................................. 109 
Table S1.3 ................................................................................................................. 110 
Table S1.4. ................................................................................................................ 111 
Table S1.5 ................................................................................................................. 112 
Table S1.6. ................................................................................................................ 113 
Table S2.1. ................................................................................................................ 115 
Table S3.1 ................................................................................................................. 118 
Table S3.2. ................................................................................................................ 119 
Table S4.1 ................................................................................................................. 125 
Table S4.2. ................................................................................................................ 126 
Table S4.3. ................................................................................................................ 127 
Table S4.4. ................................................................................................................ 128 
Table S4.5. ................................................................................................................ 129 
Table S4.6. ................................................................................................................ 130 
Table S4.7. ................................................................................................................ 131 
Table S4.8 ................................................................................................................. 132 
Table S5.1. ................................................................................................................ 136 
Table S5.2 ................................................................................................................. 137 
Table S5.3 ................................................................................................................. 138 
 
vi 
 
 
 
List of Figures 
 
Figure 1.1 .................................................................................................................... 38 
Figure 1.2 .................................................................................................................... 39 
Figure 1.3 .................................................................................................................... 40 
Figure 1.4 .................................................................................................................... 41 
Figure 2.1 .................................................................................................................... 74 
Figure 2.2 .................................................................................................................... 75 
Figure 2.3 .................................................................................................................... 76 
Figure 2.4 .................................................................................................................... 77 
Figure 3.1 .................................................................................................................. 100 
Figure 3.2 .................................................................................................................. 101 
Figure 3.3 .................................................................................................................. 102 
Figure 3.4 .................................................................................................................. 103 
 
Figure S3.1 ................................................................................................................ 120 
Figure S4.1. ............................................................................................................... 133 
Figure S5.1. ............................................................................................................... 139 
 
  
vii 
 
 
 
Overview: 
Migration, the cyclical large-scale movement of animal populations,  is an impressive 
physiological and navigational feat thought to have evolved in many species as a 
strategy to inhabit seasonal environments (Milner-Gulland et al. 2011). The large-
scale movements of migratory animals split their annual cycle into periods that occur 
in multiple ecologically and geographically distinct environments that require novel 
adaptive strategies to negotiate these disparate locations (?kesson and Hansson 
2014). The large-scale movements of migrants, however, also make them more 
susceptible to environmental change and degradation since they can potentially be 
affected at any or all of the habitats they occupy across their annual cycle (Lindstr?m 
et al. 2014).  
Recent declines in long-distance migratory songbirds that travel some of the 
longest known distances between breeding and non-breeding grounds, have brought 
into focus the need to disentangle the processes regulating these populations across 
the annual cycle (Faaborg et al. 2010). The largest gaps in our knowledge of the 
mechanisms mediating these species declines stem from the logistical difficulties in 
monitoring individuals across the annual cycle and despite migratory songbirds 
spending only 3-4 months on their breeding grounds, the vast majority of research has 
focused on this period (Marra et al. 2015). During the non-breeding season, survival 
is the key vital rate for migrants and the primary ecological drivers for this are food 
and habitat. However, factors limiting migrants can also act indirectly through non-
lethal effects that carry-over to influence survival in subsequent seasons or 
reproductive success (Marra et al. 1998, Harrison et al. 2011). Determining the key 
1 
 
 
 
ecological factors limiting demographic characteristics across the non-breeding 
periods will first require an understanding of the ecological and physiological 
strategies employed by species during this period, something lacking for the majority 
of bird species. 
The goal of this dissertation was to contribute to a better understanding of the 
ecological factors driving population dynamics during the non-breeding season of a 
declining migratory songbird, wood thrush (Hylocichla mustelina). To do this I 
explored the ecological factors driving non-breeding space-use strategies and 
movement dynamics. I employed both observational and experimental approaches to 
monitor the behavioural strategies employed by wood thrush in response to 
environmental heterogeneity both within and across the non-breeding periods. In my 
first chapter I examined how behavioral strategies vary across periods of the annual 
cycle by investigating seasonal variation in habitat selection. In the final two chapters 
I focused on the role of one ecological factor, food availability, and how it potentially 
influences the movement ecology of wood thrush during the non-breeding stationary 
period (hereafter winter) and spring migration.  
Wood thrush are large (40-50g), omnivorous Neotropical migratory songbirds 
that breed across eastern North America and winter in Mexico and Central America. 
Similar to many long-distance Neotropical migratory species, their populations have 
declined by 60% over the last 50 years (Sauer et al. 2013). Wood thrush are an ideal 
study system for a study of movement ecology because they are a common forest 
birds, they engage in multiple non-breeding space-use strategies (Rappole et al. 
2 
 
 
 
1989), are large enough to carry a range of tracking devices necessary for monitoring 
a highly mobile species, and exhibit site fidelity across their breeding and wintering 
range making it easier to recover archival tracking devices.  
Determining how migratory songbirds respond to environmental 
heterogeneity across the annual cycle will be critical for identifying habitat 
requirements across each life-history stage, information lacking for most species. In 
Chapter 1, I deployed high-resolution GPS transmitters across five breeding 
populations to quantify habitat selection at multiple spatial scale (regional, landscape, 
local) as wood thrush moved across periods of their annual cycle. For the first time 
for a migratory songbird I obtained direct evidence from individually tracked birds of 
seasonal shifts in habitat selection preferences. In general, migratory periods were 
characterized by greater behavioural flexibility and a shift to more generalist habitat 
selection. I also employed a novel technique to identify regional habitat selection 
across populations by quantifying patterns of migratory connectivity between 
breeding and non-breeding locations. The identification of regional habitat selection 
and population-specific patterns of habitat use during fall migration and the winter 
period suggest populations may exhibit different habitat selection preferences. 
Together these results suggest the factors driving the evolution of habitat selection 
preferences (e.g. food availability, timing of breeding) vary across seasons but also 
environmental context and should be considered when assessing habitat suitability for 
migratory species across the annual cycle.   
3 
 
 
 
Food availability has been identified as a key ecological factor contributing to 
non-breeding season limitation, notably in Neotropical migratory warblers (Sherry et 
al. 2005, Brown and Sherry 2006a, Smith et al. 2011a). Therefore, in the final two 
chapters I examined the drivers and consequences of differential winter food 
availability on the movement behaviour of wood thrush during the winter and spring 
migratory period.  In Chapter 2, I examined how moisture gradients drive food 
availability and in turn influence space-use strategies during the winter period. By 
monitoring birds using radio telemetry at a study site in Belize I found evidence of 
food limitation for non-breeding wood thrush that was driven by patterns of moisture. 
Dry conditions drove low food availability, low individual body condition and the 
adoption of larger home ranges. By combining radio telemetry data from our study 
site in Belize and GPS tracking from Chapter 1, I discovered that a high proportion of 
wood thrush from across the breeding range engaged in permanent large-scale mid-
winter shifts in home ranges. Wood thrush in wetter habitats had a higher probability 
of engaging in these large-scale facultative movements. I suggest that the facultative 
movements are a condition-dependent strategy to allow wood thrush to find 
alternative habitats as conditions deteriorate across the dry season. These results 
suggest that food availability driven by patterns of moisture influence both small- and 
large-scale space use strategies for non-breeding wood thrush 
In Chapter 3, I evaluated if differential non-breeding food availability could 
potentially influence the endogenous spring migratory schedules of wood thrush by 
performing a food manipulation study on a captive population. By holding birds 
under two feeding treatments (ad libitum or ?control? vs. reduced food availability 
4 
 
 
 
expected to result in a 5% decline in body weight) I found that food availability and 
body condition modulated the intensity of migratory restlessness (Zugunruhe), an 
index of migratory disposition in captive birds. Low food availability did not 
influence the onset of Zugunruhe, but migratory fattening was advanced. The 
advancement of migratory fattening when food availability is low may provide a 
mechanism to flexibly adjust migration timing under poor environmental conditions. 
These results suggest that non-breeding food limitations could constrain migration 
preparation in wood thrush and highlights the tricky balancing act faced by wood 
thrush between optimizing survival during migration and maintaining optimal spring 
migration schedules.  
The results described here and in more detail in the following chapters 
identify a diversity of behavioral strategies employed by wood thrush to handle 
environmental heterogeneity during the non-breeding periods. They suggest 
flexibility in space-use strategies and movement dynamics both across and within 
seasons as individuals adjust to different life-history stages or environmental 
conditions. This research also highlights the importance of considering the full annual 
cycle of migratory songbirds, including seasonal interactions, when investigating 
factors limiting migratory species.  As environmental degradation and climate change 
proceeds understanding how migratory birds respond to environmental heterogeneity 
will be increasingly important for the management of declining species. Across the 
non-breeding range of wood thrush conditions are projected to become increasingly 
arid and the research presented here suggests that these climate shifts could have 
important consequences for wood thrush both within the non-breeding periods and 
5 
 
 
 
potentially into subsequent periods through seasonal interactions in their annual life 
cycle.  
6 
 
 
 
  Year-round habitat selection for a Neotropical 
migratory songbird using high resolution GPS tags 
Abstract 
Deciphering how environmental heterogeneity affects population dynamics in 
migratory species is complicated by the redistribution of individuals in time and space 
across the annual cycle. Approaches that tackle this problem require information 
about how migratory species respond to ecological factors across time and space, and 
how they are linked across migratory periods.  Using high spatial resolution GPS 
tracking of individual male songbirds, we quantified for the first time (1) local and 
landscape-scale habitat selection across the annual cycle and (2) patterns indicative of 
regional habitat selection for five populations of wood thrush (Hylocichla mustelina) 
throughout their breeding range. We found that wood thrush exhibited seasonal 
variation in habitat selection. Across stationary periods, local habitat selection 
included variables associated with forest habitat, notably, a preference for forest edge 
habitat. In contrast, during migratory periods wood thrush exhibited greater 
behavioural flexibility indicative of a more generalist approach to habitat selection. 
Landscape habitat selection was only identified during the breeding season (average 
patch size) and could be a result of extensive forest fragmentation in the North 
American breeding grounds. We also identified individual distribution patterns 
indicative of regional habitat selection during fall migration and the stationary non-
breeding period, but not spring migration. In accordance with patterns of regional 
habitat selection, we found population-specific patterns in habitat use (e.g. distance to 
forest edge). Seasonal changes in habitat selection at multiple spatial scales suggest 
7 
 
 
 
the factors driving habitat selection patterns are aligned with life-history stage and 
may be dependent on regional differences in landscape composition. These results 
highlight the importance of a full annual cycle approach to ecological studies that 
address how migratory species respond to spatial and temporal environmental 
heterogeneity.  
 
Introduction  
The annual cycle for a migratory species is characterized by long-distance movement 
of individuals between multiple geographically, and often ecologically, disparate 
locations (Newton 2008, ?kesson and Hansson 2014). In each location, 
environmental habitat heterogeneity drives individual settlement decisions and habitat 
selection processes often have individual fitness consequences (Hutto 1985, Mayor et 
al. 2009). Migratory birds, despite their small size, travel some of the longest 
distances of any animal (Alerstam et al. 2003). Although empirical studies of habitat 
selection have been conducted within breeding (Orians and Wittenberger 1991, 
Mitchell et al. 2001, Lee et al. 2002), migration (Buler et al. 2007, Mccabe and Olsen 
2015, Lafleur et al. 2016) and non-breeding stationary periods (Chandler and King 
2011, Fraser et al. 2017), few species have been assessed across multiple seasons. 
Moreover, studies rarely use the same individuals or populations (but see Beatty et al. 
2014, Pickens et al. 2017). As such, we know relatively little about the breadth of 
suitable habitats and how individuals select habitats throughout the annual cycle for 
migratory birds (Marra et al. 2015, McGarigal et al. 2016).  
8 
 
 
 
Adaptive habitat selection has been theorized as a hierarchical decision-
making process by which the factors that drive habitat suitability (based on fitness 
costs and benefits) and the mechanisms for assessing suitability occur at different 
spatial and temporal scales across the annual cycle (Figure 1.1; Johnson 1980, Wiens 
1989, Mayor et al. 2009).  At each spatial or temporal scale, the decision making 
process may involve novel and multiple interacting criteria (e.g., patch size, predation 
rate) and be constrained by the effects of criteria from higher scales (annual vs diel; 
Rettie and Messier 2000, Mayor et al. 2009). When considering spatial scales, habitat 
selection decisions at the broadest scale will shape the species' geographic range and 
patterns of regional population distribution (hereafter referred to as ?regional?; 
Johnson 1980, Meyer and Thuiller 2006). Hutto (1985) refined this idea for migratory 
birds and suggested non-habitat factors are expected to drive decisions at the regional 
scale. For example, an individual?s genetically programmed and learned migration 
route (Hutto 1985), weather (Moore and Aborn 2000, Buler and Moore 2011), or 
physiological condition (Studds et al. 2008, Rushing et al. 2015) are factors expected 
to play a role in regional habitat selection.  In contrast, as the spatial scale decreases, 
Hutto (1985) suggested that habitat selection will increasingly be driven by the costs 
and benefits associated with the habitat itself. At intermediate scales (hereafter 
referred to as ?landscape?), selection drives the choice of a particular habitat type and 
home range. Finally, at finer spatial scales, selection of habitats within the home 
range (hereafter referred to as ?local?) or microhabitat selection is most closely linked 
to the ultimate factors driving habitat suitability (e.g. food or nesting site availability). 
Equally important to consider is how the decision-making process changes across 
9 
 
 
 
temporal scales or levels (Cody 1985, Wiens 1989). For example, factors limiting an 
individual?s fitness can vary between diel and seasonal decisions (scales) or equally 
across different seasons (levels; Mayor et al. 2009). Therefore, habitat selection may 
be a hierarchical process that operates across multiple spatial scales ranging from 
regional to micro-habitat but also across temporal scales.   
Over the last thirty years, the multi-scale approach to habitat selection has 
been incorporated into many ecological studies across a range of vertebrate taxa 
(Wiens and Milne 1989, Orians and Wittenberger 1991, Rettie and Messier 2000, 
Grand and Cushman 2003). For instance, Buler et al. (2007) found that factors driving 
the selection of stopover habitat by migrant songbirds along the Gulf of Mexico 
differed across multiple spatial scales. At regional spatial scales the most relevant 
limiting factors driving habitat selection were weather and physiological condition 
(non-habitat factors) after crossing a large ecological barrier; whereas, at the 
landscape and microhabitat scale the limiting factor was food (habitat factor). Overall 
many studies have successfully shown that habitat selection is scale-dependent, 
largely in terms of spatial scales; however, the inferences that can be drawn from 
these studies have often been limited due to the low spatiotemporal extent at which 
they were conducted (McGarigal et al. 2016).  
For migratory songbirds, most ecological studies have been conducted during 
the breeding season (Marra et al. 2015). Although habitat selection studies are 
increasingly conducted across multiple scales (landscape to microhabitat, seasonal to 
diel), their scope has been restricted to studies of single seasons on single 
populations. When habitat use across seasons has been documented, many migratory 
10 
 
 
 
songbirds, even those considered habitat specialists, appear to show shifts in habitat 
use during the non-breeding season (Petit 2000, Zuckerberg et al. 2016). This lack of 
consistency in habitat associations suggests that individuals can maximize their 
fitness across seasons through behavioural flexibility in their habitat settlement 
decision-making. (i.e. across life-history stages: nesting vs. re-fueling). However, 
documenting behavioral flexibility in habitat selection has remained challenging due 
to the difficulty in tracking small mobile animals across large spatial and temporal 
scales. Therefore, two large information gaps related to behavioral flexibility in 
habitat selection need to be addressed: 1) what factors drive habitat selection 
decisions across seasons for individual birds, and 2) what factors underlying habitat 
selection vary across the species range? 
In migratory species, regional habitat settlement decisions occur seasonally. 
The practical limitations of sampling across large spatiotemporal scales has prevented 
studies of regional habitat selection and the ultimate factors (extrinsic vs. intrinsic) 
and cues (innate vs. learned) driving selection at this scale (Figure 1.1; Moore et al. 
1993, Kelly et al. 1999, McGarigal et al. 2016). Migratory connectivity is most 
commonly used to describe the strength of regional re-distribution of migratory 
individuals across the annual cycle, but studies of migratory connectivity could also 
provide important insights into the non-habitat factors that affect regional habitat 
selection patterns. For example, patterns of migratory connectivity could be used to 
identify the presence of population-specific regional habitat selection decisions as 
individuals move across large distance throughout the year. Due to the hierarchical 
nature of the habitat selection process migratory connectivity can also provide a 
11 
 
 
 
measure of how regional habitat selection might be expected to constrain settlement 
decisions at finer spatiotemporal scales. Given that the environment is expected to 
vary across a species range, better integration of regional scale habitat selection could 
have important implications for our interpretation of habitat selection patterns across 
a species range.  
Here, we evaluated the spatial and temporal patterns of habitat selection at 
three spatial scales (local, landscape and regional) and across all seasons (breeding, 
fall migration, winter and spring migration) of the annual cycle for wood thrush 
(Hylocichla mustelina), a Neotropical migratory songbird using fine-spatial resolution 
archival GPS geolocators deployed across five distinct breeding populations. First, to 
determine how factors driving habitat selection change across seasons, we explored 
individual variation in habitat selection at the local and landscape scale. Since wood 
thrush are characterised as forest interior species that are tolerant of forest 
fragmentation (Evans et al. 2011), we predicted that if habitat selection remains 
constant across seasons, thrushes will consistently select local and landscape features 
characteristic of interior forest habitats. In contrast, if habitat selection exhibits 
seasonal plasticity, we expected habitat selection to differ during migratory periods 
leading to more generalist habitat selection as birds become more responsive to non-
habitat features (e.g. temporal constraints on migration, energetic demands) and have 
more difficulty accurately assessing the value of unfamiliar habitats.  In addition, we 
predicted that wood thrush will exhibit local and landscape scale habitat selection 
during stationary (breeding and winter) periods, but selection will shift to landscape 
12 
 
 
 
scale features during migration, as they provide a quick cue for birds to assess habitat 
quality (Buler et al. 2007, Beatty et al. 2014).    
Second, to test for seasonal changes in regional habitat selection between 
breeding populations we assessed the strength of migratory connectivity.  Based on 
earlier work on wood thrush migratory connectivity with low resolution tracking 
devices (Stanley et al. 2015), we predicted the strength of regional habitat selection 
would vary seasonally. Specifically, strong connectivity during the fall is driven by 
population specific route (Stanley et al. 2015); these patterns predict regional habitat 
selection during fall stopover at the Gulf of Mexico. By contrast, from breeding to 
winter sites, wood thrush show moderate connectivity (Stanley et al. 2015); these 
patterns suggest weaker patterns of regional habitat selection. Finally, weak 
connectivity during spring migration is driven by convergence of migratory routes at 
the Gulf of Mexico; these patterns suggest no regional habitat selection.   
Third, if regional habitat selection results in breeding populations remaining 
geographically isolated across the annual cycle, we predicted populations will diverge 
in habitat selection at finer spatial scales.  Correspondingly, we expected individuals 
from the same breeding populations to have similar patterns of habitat use during fall 
migration and the winter period, when regional habitat selection is predicted to occur, 
but no differences in habitat use between breeding populations during spring 
migration when regional habitat selection is not predicted to occur.  
13 
 
 
 
Methods 
Study system  
Wood thrush are medium sized, long-distance Neotropical migratory songbirds that 
breed in deciduous and mixed forests of eastern North America and winter in the 
broad-leaf and palm forests ranging from southern Mexico to northern Panama 
(Evans et al. 2011). They are a multi-brooded, omnivorous species that primarily 
forages on the ground. Wood thrush maintain territories during the winter and 
breeding season; however, they show extensive flexibility in territorial behavior. 
Tracking studies have found they engage in regular off-territory forays and can make 
large scale relocations (1-25 km) throughout stationary periods (Rappole et al. 1989, 
Lang et al. 2002). Their global population size has declined by 60% since 1966 based 
on breeding ground surveys (Sauer et al. 2013). 
Field Methods and GPS telemetry 
To determine spatio-temporal variation in habitat selection across the annual cycle we 
studied wood thrush from 2014-2015. Archival GPS tags (Model PinPoint-50, 1.8 g, 
50 fixes, Lotek Wireless) were deployed on 137 breeding adult male wood thrush 
over the summer of 2014 and 2015 in Delaware, Indiana, North Carolina, New York 
and Minnesota (Appendix S1, Supplementary Methods, Table S1.1). The tags provide 
location estimates for individual birds with an estimated accuracy of 10 meters. GPS 
tags were recovered by returning to sites the following summer and recapturing 
individuals. We recovered 23 tags, and retrieved data from 21 tags. Two tags were 
unresponsive (no data retrieved) and 9 lost antennas sometime after release (partial 
14 
 
 
 
data recovery). We obtained an average of 26 out of a maximum of 50 points per tag 
(range: 1-46; Appendix S1, Table S1.2). Most missing fixes, excluding those 
involving lost antennas, occurred during migratory periods (Appendix S1, Table 
S1.2). Due to low success of location fixes during migratory periods, annual cycle 
stages (breeding, fall migration, winter, spring migration) were defined based on the 
first or last location recorded at stationary stages (i.e. spring migration ends day 
before first recorded point on breeding grounds). Migratory periods were identified as 
northward or southward movements over the course of consecutive days (+2 points, 6 
- 8 days) and larger than 20km. 
Habitat Features: 
To quantify the structural characteristics of habitats occupied by wood thrush across 
the annual cycle we paired location fixes to geospatial habitat data. Location 
estimates spanned eastern North America and Central America; therefore, we 
restricted geospatial data to those sources that covered the entire study area (detailed 
description of satellite imagery in Appendix S1, Table S1.3). At each location fix we 
examined habitat use characteristics at two spatial scales (local and landscape) using 
the extract function in the R package raster (Hijmans 2016) . Local habitat features 
represented remotely sensed data interpolated at each point location and included 
percent tree cover, Enhanced Vegetation Index (EVI), patch size (km2) and proximity 
to non-forest (m; detailed description of interpolation methods Appendix S1, Table 
S1.3). Landscape characteristics represented remotely sensed data interpolated from a 
buffer of 1 km radius around each point location and included average percent tree 
cover and average patch size (km2). A 1 km buffer was used because off-territory 
15 
 
 
 
forays of wood thrush typically range from 150 m to 1 km (Rappole et al. 1989, Lang 
et al. 2002). To assess landscape composition, we interpolated data from classified 
land cover layers using a 5 km buffer (Appendix S1, Table S1.3). A 5 km buffer was 
used because classified land cover layers suitable for a 1 km buffer were unavailable 
across our study area, landscape composition was therefore analyzed separately from 
the other landscape characteristics.  
Local and Landscape Habitat Selection 
To identify variation in spatio-temporal patterns of habitat selection at the individual 
level we explored three measures. First, at the landscape scale, we examined diversity 
in landscape composition across seasons.  Second, we quantified the individual 
repeatability of habitat features at both the landscape and local scales across the 
stationary (breeding and winter) and migratory periods (fall and spring migration). 
Finally, to disentangle whether seasonal differences in habitat usage were driven by 
changes in habitat availability across regions or shifts in factors driving habitat 
selection, we employed a multi-scale approach to determine habitat selection within a 
use-availability design.  
First, to estimate seasonal diversity in landscape composition for each 
individual we calculated the Shannon diversity index (Shannon 1948). The index was 
assessed each season based on landscape composition (the number of classified land 
cover classes) at each point location used by an individual with the R package vegan 
(Oksanen et al. 2017). High landscape diversity indicates individuals? occupied 
landscape with multiple habitat types across the season. 
16 
 
 
 
Second, for each habitat feature we calculated the repeatability of use among 
individual birds between stationary and migratory periods. Repeatability quantifies 
the proportion of between-individual variation from observations relative to the total 
variation in the population for a repeated measure (Lessells and Boag 1987). 
Repeatability increases when within-individual variation is low (e.g. individuals 
chose locations with consistent habitat features) and between-individual variation 
increases (individuals chose different features). To calculate repeatability we used the 
r package rpt (Stoffel et al. 2017) based on the structure of the best-fit model 
describing each habitat feature (see Appendix S2).  
Third, we constructed discrete-choice models for each season to determine 
seasonal habitat selection preferences of wood thrush. These models examine the 
probability of an individual choosing a location based on a choice set of alternative 
available locations (Cooper and Millspaugh 1999). If habitat features are used 
disproportionate to their availability, it is assumed they confer a fitness advantage 
(Manly et al. 2004). If these habitat features change across seasons, we conclude this 
represents behavioural flexibility in habitat preferences. GPS location fixes were 
considered as used resource units. Available resource units were generated by 
creating a buffer around each used resource unit and generating up to 20 random 
points. Therefore, each used resource unit had its own set of available units. Due to 
varying levels of cloud coverage across the study area, number of available points 
ranged from 10 to 20. Consistent parameter estimates can be obtained with as few as 
four available alternatives (McFadden 1978); however, to account for spatial 
variation in habitat features we used up to 20 points when possible (Bonnot et al. 
17 
 
 
 
2011).  For all used and available points, local and landscape habitat features were 
extracted as defined in the habitat features section. Fewer location estimates were 
obtained from each individual during migratory periods (mean number location fixes, 
migration = 4.3; stationary = 10.25). To account for different numbers of replicate 
observations during migratory periods, only 4-5 location fixes were used from the 
stationary periods for each individual (Appendix S1, Table S1.2). Points were 
subsampled within stationary periods by selecting used resource units with the largest 
distance between them. 
A multi-scale approach was used to identify habitat selection patterns that 
would be ecologically relevant to wood thrush (Wiens 1989, Buler et al. 2007). Since 
relevant spatial extents were unknown for wood thrush (size of relevant study area), 
we delineated four spatial extents encompassing distances moved within an 
individual?s territory (0.5 km), during off-territory forays (5km), and during 
relocation events (15 and 25 km) based on prior information on wood thrush spatial 
ecology. The spatial extent defined the size of the buffer created around each used 
resource point and served to generate the available points for each used point. 
Bayesian mixed conditional logistic regressions were used to model the probability of 
a wood thrush choosing a used resource unit out of the available choice set. The 
mixed conditional logistic regression was adapted from the methods of Beatty et al. 
(2014) and a Bayesian framework was employed because comparable mixed 
conditional logistic regression were not available in existing Maximum Likelihood R 
packages. We developed a set of candidate models to test alternative hypotheses 
about the habitat features influencing habitat selection (Appendix S1, Table S1.4). All 
18 
 
 
 
models, with the exception of the null model, included individual bird as a random 
variable to account for variation in selection patterns among birds. At the 0.5-km and 
5-km spatial extent we evaluated two models (local and null) and at the relocation 
level (15-km and 25-km) we evaluated four models (local, landscape, full, and null). 
These models were run for each season for a total of 48 models (12 models x 4 
seasons).  
Regional Habitat Selection 
To identify regional habitat selection, we estimated migratory connectivity between 
(1) breeding and wintering grounds and (2) breeding and northern Gulf of Mexico 
stopover sites. We employed two connectivity metrics: (i) the Mantel correlation 
using the ade4 package in R  (Dray and Dufour 2007) and (ii) the MC metric from the 
MigConnectivity package (Cohen et al. 2018) using orthodromic distances between 
breeding and winter or stopover sites (detailed methods in Appendix S1). To examine 
the consistency of annual schedules between breeding populations, we determined the 
timing of annual events for all individuals (see Appendix S3).  
If regional habitat selection leads to population-specific habitat selection we 
expected to see population-specific variation in habitat use patterns (description of 
used habitat features) across the annual cycle. Due to limited sample size, we were 
unable to assess population-specific habitat selection (the disproportionate use of 
habitat components). Instead, we described population-specific patterns of local and 
landscape scale habitat features (characteristics of used habitats). To model 
population-specific habitat use we used a generalized linear model for each habitat 
feature (local and landscape). The full model included the habitat variable as the 
19 
 
 
 
response variable, the interaction of season and breeding longitude as fixed effects 
and individual bird as a random effect. 
Statistical Analysis 
All analyses were conducted in R 3.4.3 (R Core Team 2017). Bayesian discrete 
choice models were run with the R package jagsUI (Kellner 2017) using the software 
JAGS 4.3.0 (Appendix S1, Plummer 2003). Generalized mixed effect models were 
run in the R package nlme (Pinheiro et al. 2017) with restricted maximum likelihood 
estimation. Generalized linear models were run in the R package stats (R Core Team 
2017). For full model specifications and fit assessment see Appendix S1. 
Results 
Patterns of Movement  
We obtained breeding ground location estimates from 21 birds, fall migration 
locations from 20 birds, winter locations from 19 birds (1 partial) and spring 
migration from 18 birds for a total of 554 GPS location fixes across 21 birds (Figure 
1.2; Appendix S1, Table S1.2). During the winter period, of the 18 birds tracked 
across the whole period, 8 individuals engaged in intra-winter movements occupying 
two or more territories an average of 58.5 ? 20.8 km (? SE, range: 1.5 ? 182.3 km) 
apart (Figure 1.2c). The remaining 10 individuals remained at a single winter 
territory, moving an average of 0.1 ? 0.09 km between locations from November to 
April.  
 
20 
 
 
 
Local and Landscape Habitat Selection 
The diversity of land cover type occupied by individual wood thrush varied by season 
(?2 = 45.92, df = 3, p < 0.001; Figure 1.3). Residual plots for the top model showed 
deviations from model assumptions so non-parametric Kruskal-Wallis and Dunn 
Tests with Bonferroni adjustments were used.  We did not detect a difference in land 
cover diversity within stationary or migratory periods (Breeding-Winter: Z = -0.26, p 
= 1.0), but we did detect a difference in land cover diversity between the migratory 
and stationary periods (Breeding-Fall: Z = 5.79, p < 0.001; Breeding-Spring: Z = -
3.47, p = 0.003; Winter-Fall: Z = 5.46, p < 0.001, Winter-Spring: Z = 3.18, p = 0.08). 
Within the breeding season the major land cover type used was a matrix of mixed-use 
agriculture and natural vegetation, while the dominant land cover type on the winter 
grounds was forest (Figure 1.3). The high diversity of land cover types selected by 
wood thrush during fall and spring migration included forest, agriculture, agricultural 
mosaics, wetlands and wooded savannahs.  
We quantified the repeatability of selection of habitat features by individual 
birds as they moved within the stationary and migratory periods. Within the 
stationary periods, repeatability of local and landscape tree cover, proximity to non-
forest and average patch size was moderate (R > 0.3; range, R = 0.327 ? 0.483, Table 
1.1). EVI and patch size had low repeatability (R < 0.3) during stationary periods. By 
contrast, all habitat features had very low repeatability estimates, close to zero, within 
the migratory periods.  
Discrete choice models indicated that the habitat features of the resource units 
(locations) wood thrush selected (compared to what was available) varied across 
21 
 
 
 
seasons and spatial extent. The top discrete choice model of habitat selection for most 
spatial extents during the breeding, winter and fall migratory period was the local 
habitat features model (Figure 1.4; Appendix S1, Table S1.5). The exception was 
during the breeding season at the 25-km spatial extent, in which the full model (local 
and landscape habitat features) was the top model. During spring migration, the null 
model was the top model across all scales, indicating no habitat features measured 
were important for habitat selection.  
On the breeding grounds at all spatial extents, birds selected resource units 
with high percent tree cover (Figure 1.4). At the smallest spatial extent (0.5 km) birds 
chose resource units with low EVI values.  At the two largest spatial extents (15-km 
and 25-km), wood thrush selected resource units more proximate to non-forest. 
Additionally, on the breeding grounds, at the 25-km extent birds selected resource 
units that were found within landscapes with high average patch size (1-km buffer), 
but at the local scale, resource units in small forest patches were selected. During fall 
migration, birds selected resource units with large patch size at the 0.5-km, 5-km and 
25-km spatial extent. At the two largest spatial extents (15-km and 25-km) they chose 
resource units with high EVI values. On wintering grounds, at all spatial extents birds 
chose resource units with large patch sizes. Birds also chose winter resource units 
more proximate to non-forest at the 5, 15 and 25-km spatial extent. Additionally, at 
the 25-km scale birds chose resource units with high tree cover.  
Regional Habitat Selection  
Between breeding and stationary non-breeding grounds regional habitat selection was 
present, and migratory connectivity estimates were weak-to-moderate (MC Index = 
22 
 
 
 
0.41; rM = 0.14, p = 0.03). Regional habitat selection was also present at fall 
stopovers locations prior to the Gulf of Mexico crossing based on high migratory 
connectivity estimates (MC index = 0.69; rM = 0.69, 9 < 0.01) due to breeding 
populations utilizing region-specific departure points from North America (Figure 
1.2b). By contrast, during spring migration, connectivity was low (MC index = 0.28; 
rm = -0.03, p = 0.53) at stopovers after Gulf of Mexico crossing suggesting no 
regional habitat selection by breeding populations.  
There was evidence of population-specific patterns of seasonal habitat use 
based on two local habitat features (patch size and proximity to non-forest), and one 
landscape level characteristics (average patch size; Table 1.2 and Appendix S1, Table 
S1.6). There was no difference in land cover diversity between populations. During 
the breeding season, eastern populations were found in locations with significantly 
higher patch size compared to western populations, whereas western populations 
were found in locations significantly further from non-forest compared to eastern 
populations (Table 1.2). During fall migration, proximity to non-forest at fall stopover 
sites was significantly greater for eastern breeding populations compared to western 
populations. During the non-breeding stationary period, western populations were 
further from non-forest and were located in landscapes with higher average patch 
size. There were no significant trends between breeding longitude and habitat features 
during spring migration.  The model of average tree cover included a significant 
interaction between breeding longitude and season, but the confidence intervals of all 
trends included zero (Table 1.2).  
23 
 
 
 
Discussion 
For the first time in a migratory songbird we show direct evidence, from individual 
birds with high resolution tracking devices, of seasonal shifts in habitat selection 
across the annual cycle. Most notably individual birds relaxed the degree to which 
they selected habitat features during migratory periods, suggesting greater 
behavioural plasticity during migration. We also show evidence of regional habitat 
selection by using indices of migratory connectivity to assess non-random assortment 
of five populations across regional geographic scales. As migratory species move 
through different heterogeneous landscapes they will need to find suitable habitats 
that meet their needs during different life-history stages. Understanding the factors 
necessary for persistence will require identifying the breadth of suitable habitats that 
individuals occupy as they redistribute geographically across the annual cycle 
(Webster and Marra 2005).  
Seasonal Variation in Habitat Selection 
This is the first study we are aware of that has documented seasonal variation in 
habitat selection by tracking the same individual migratory songbird across the entire 
annual cycle. Fine scale tracking of individual wood thrush revealed variation in 
habitat selection across periods of the annual cycle. Specifically, by employing a 
multi-scale approach we determined wood thrush were selective at the local scale and 
only selective at the landscape scale during the breeding season. The factors that 
drove habitat selection varied across seasons and spatial extent (size of study area). 
Habitat selection by migrating wood thrush was less selective, individuals occupied a 
24 
 
 
 
greater diversity of land cover types than during the stationary periods, and selection 
focused on local habitat features. 
During the stationary non-breeding and breeding periods we found individuals 
almost exclusively in forested and mixed forest-agriculture landscape. Alternatively, 
while migrating, wood thrush selected novel land cover types (e.g. agricultural, wood 
savannah) 36% - 46% of the time. Furthermore, during migration the low 
repeatability in habitat use indicated that individual birds did not consistently select 
locations with similar structural attributes. Banding and eBird studies have also 
suggested greater flexibility in habitat selection for songbirds during the migratory 
period and changes in foraging behaviour, particularly for fruit-eating species like 
wood thrush (Yong and Moore 2005, Zuckerberg et al. 2016). Analyses were not 
conducted on the same individuals or population; however, this study confirms that 
changes in habitat use during the migration period were due to changes in individual 
habitat selection across periods. Specifically, unlike the stationary periods, we found 
fewer predictors of habitat selection during fall migration and no response to habitat 
features during spring migration, which suggests birds were selecting habitats at 
random with regards to the features we measured during spring migration.  
This shift to be more of a generalist when selecting habitat may be favoured 
due to the strong selection pressure to minimize time on migration (Hutto 1985, 
Alerstam and Lindstr?m 1990, G?mez et al. 2017). This is expected to be particularly 
relevant during spring migration due to intense selective pressure for early arrival on 
breeding grounds to secure a territory and begin reproduction (Kokko 1999). 
Migration schedules should therefore ensure optimal arrival dates on breeding 
25 
 
 
 
grounds, which is consistent with our finding of faster spring migrations and 
population-specific schedules (Appendix S3, Figure S3.1, Table S3.2).  Increased 
behavioural flexibility during migration may also be an adaptive response to lower 
migrants? threshold of acceptable habitat to reduce search time for high quality 
stopover sites (Moore and Aborn 2000). Therefore, if individuals cannot quickly 
locate high quality habitat they may settle in poorer quality habitats. It has been 
suggested that compensatory behaviours such as risk-prone foraging strategies (e.g. 
increased maneuvers or feeding intensity) or shorter stopover duration, may allow 
birds to maintain migration schedules even when foraging in poorer quality habitats 
(Yong and Moore 2005, Nilsson et al. 2013). Therefore, during spring migration, at 
fine spatial and temporal scales food may still be limiting individuals, but at larger 
spatial scales the timing of migration may be the more limiting factor.  
During fall migration, wood thrush selected larger forest fragments and, when 
examined across large spatial extents (15, 25 km), patches with higher EVI values. In 
contrast to spring migration, birds were responding to local habitat features, but to a 
lesser extent compared with the stationary periods. We found fall migration schedules 
were more variable across individuals and populations and longer in duration 
(Appendix S3 Figure S3.1, Table S3.2), which suggests individuals may fine tune fall 
migration to adjust to local conditions en route (Balbont?n et al. 2009, Stanley et al. 
2012). Settlement patterns during migrations are thought to be driven largely by food 
to meet the energetic demands of migration and this is consistent with the response of 
birds to EVI, an indicator of primary productivity (Hutto 1985, Buler et al. 2007). 
Behavioural plasticity between migratory periods could be driven by seasonal 
26 
 
 
 
difference in environmental conditions linked to food availability. Increased food 
limitation in the fall, driven by poorer seasonal conditions en route or carrying-over 
from the breeding/moulting period, could lead to increased selection for habitat 
preferences (Nilsson et al. 2013). In contrast to spring migration, timing of arrival on 
winter grounds may not be associated with fitness advantages, which could drive 
differences in behavioural plasticity between seasons (M?ller 1994, McKellar et al. 
2013b).  
Landscape scale habitat selection was only identified during the breeding 
season across the largest spatial extent. The majority of previous studies have also 
found that local, and not landscape, habitat features better predicted wood thrush 
occupancy during both stationary periods (breeding, Lee et al. 2002, Valente and 
Betts 2018; non-breeding, Graham and Blake 2001). The relevance of landscape 
variables during the breeding season, but no other season could be driven by several 
factors. It could be that critical resources unique to the breeding season (e.g. 
reproduction-related) are influenced at the landscape scale (Hutto 1985). 
Alternatively, it has been suggested that the impacts of fragmentation are primarily 
visible when habitat loss across the landscape is high or moderate (fragmentation 
threshold; Andren 1994, Villard and Metzger 2014). The dominant land cover during 
the breeding season was mixed forest-agriculture, and there was low forest cover 
across the landscape (Appendix S2, Table S2.1). In comparison, during the winter 
period, forest was the dominant land cover and forest cover at the landscape scale was 
high. Therefore, the higher fragmentation in locations occupied by wood thrush 
27 
 
 
 
during the breeding season could be driving the greater importance of landscape 
configuration in our models.  
Structural features of the locations occupied by individual wood thrush during 
the two stationary periods (non-breeding and breeding) showed moderate to high 
repeatability, once seasonal differences were accounted for. This was particularly 
notable for distance to edge; individuals that tended to be closer to edge on the 
breeding grounds also tended to be closer to edge during the non-breeding stationary 
period. This pattern was not true for patch size or EVI, which showed low 
repeatability between stationary periods. Overall, three local habitat features - tree 
cover, patch size and proximity to non-forest - were top predictors of habitat use 
during both stationary periods however their relevance varied by spatial extent and 
season. Together these results indicate that factors related to forest structure play a 
role in habitat selection during the stationary periods and suggests that individual 
birds may show distinct preferences for particular local and landscape features at 
different periods of the annual cycle.  
Breeding wood thrush settled in landscapes with large forest patches, a pattern 
that only emerged at the largest spatial extent. Within these landscapes, selection of 
sites was highly variable (demonstrated by large confidence intervals), but individuals 
more often settled within small forest fragments and in locations closer to edge 
habitat with high tree cover. It is well documented that wood thrush inhabit small 
forest patches across most of their breeding range (Robinson et al. 1995, Burke and 
Nol 2000, Lee et al. 2002), but occupancy of small forest fragments is usually 
associated with decreased reproductive success for wood thrush (Donovan et al. 1995, 
28 
 
 
 
Burke and Nol 2000). Although in some landscapes small fragments can function as 
source populations (Weinberg and Roth 1998, Fauth 2001). Individuals may be 
driven into smaller forest fragments due to selection of habitat features associated 
with small fragments (e.g. high shrub density; Cody 1985). Wood thrush also 
responded to edge habitat, which would be more common in small fragments and has 
previously been identified as an important nesting and fledgling habitat for wood 
thrush (Vega Rivera et al. 1998, Kaiser and Lindell 2007). Alternatively the pattern 
could be driven by limited high quality forest patches and the displacement of 
individual to lower quality patches through intraspecific competition (Fretwell and 
Lucas 1969, Thompson III et al. 2002). Finally, our choice of study plots and 
deployment location may have also skewed the results towards smaller fragments. 
Further research measuring fitness components would be better able to dissect which 
landscape or patch level factors increase the conservation value of small fragments.  
In contrast to the breeding grounds, the dominant land cover during the winter 
period was forest. As predicted, individuals were more likely to settle in large forest 
patches and locations with high tree cover. However, similar to the breeding grounds, 
wood thrush disproportionately selected winter locations near edge habitat. Studies of 
wintering wood thrush suggest they are forest area sensitive but are tolerant of edge 
(Graham and Blake 2001, Roberts 2011). Fruit is a patchily distributed, ephemeral 
resource often associated with edge habitats in the tropics (Lindell et al. 2007). 
Access to fruiting resources could be critical for wintering wood thrush and radio-
tracking studies have observed wood thrush making off-territory forays to large 
fruiting trees (C.Q. Stanley pers. comm.). The observed differences in tree cover and 
29 
 
 
 
patch size across stationary periods could indicate shifts in the importance of these 
critical resources. It could also represent trade-offs in preferred habitat features 
imposed by the hierarchical process of habitat selection in regions with different 
landscape composition and configuration (Hutto 1985). 
 Regional Habitat Selection 
Due to the hierarchical nature of habitat selection by birds, environmental 
heterogeneity across a species range will influence which habitat features are limiting 
populations based on their regional context. Using migratory connectivity, we 
identified non-random assortment of populations at regional scales between breeding 
and non-breeding locations, which we interpreted as evidence of regional habitat 
selection.  As predicted based on previous low-resolution tracking studies (Stanley et 
al. 2015), wood thrush exhibited regional habitat selection on their non-breeding 
stationary (MC = 0.41) and fall Gulf of Mexico stopover sites (MC = 0.69), but not 
on their spring Gulf of Mexico stopover sites (MC = 0.28).  Weak and moderate 
levels of migratory connectivity appear to be common in forest-dwelling migratory 
songbirds (migration, Kole?ek et al. 2016, Knight et al. 2018; stationary, Gilroy et al. 
2016, Finch et al. 2017) which suggests regional habitat selection occurs across many 
species. Using migratory connectivity to identify patterns of regional habitat selection 
does not elucidate the factors driving habitat selection at this scale. Hutto (1985) 
suggested they are largely factors not related to the habitat itself, for example 
physiological condition influencing regional settlement patterns on the ground 
(Studds et al. 2008, Rushing et al. 2015). Ultimately identifying what factors drive 
patterns of regional habitat selection and how flexible they are will have important 
30 
 
 
 
implications for understanding how populations will be able to respond to changing 
landscapes. 
If regional habitat selection does occur, we could expect to see population-
specific patterns of habitat selection at smaller spatial scales. Despite relatively small 
sample sizes within each breeding population, we identified population specific 
differences in habitat use patterns during the stationary periods and fall migration, but 
not spring migration as predicted based on our analysis of regional habitat selection. 
The habitat features of breeding populations were consistent with regional patterns of 
forest fragmentation across the United States (higher fragmentation in the mid-west; 
Riitters et al. 2002); populations in the mid-west were found in smaller forest 
fragments. A similar trend was also observed during fall migration which was 
characterized by region-specific migratory routes to the Gulf of Mexico; mid-western 
population were in locations closer to edge habitat. During spring migration, all 
populations funneled into the same region of the US after the Gulf of Mexico crossing 
(eastern Texas to Mississippi) resulting in all populations occupying habitats with 
similar features. High intra-specific competition at migratory bottlenecks could force 
individuals into non-optimal habitats and might have contributed to our inability to 
detect local and landscape habitat selection during spring migration. Due to limited 
sample size this analysis does not determine if the observed patterns were driven by 
populations differentially selecting locations based on these structural aspects 
(disproportionate to their ability) or by differences in the distribution of 
environmental features across the species range. However, one structural feature, 
distance to edge, had the same regional pattern across both stationary periods; mid-
31 
 
 
 
western populations selected locations further from forest edge. In addition, this 
structural characteristic had high individual repeatability across both stationary 
periods. These results hint at the possibility that mid-western populations may have a 
preference for more interior forest areas. If populations develop unique habitat 
selection preferences, either learned or innate, ignoring information about regional 
habitat selection across the annual cycle could obstruct interpretation of habitat 
selection patterns at lower spatial scale and our understanding of the conservation 
value of different habitat features across populations.  
Conservation Implications 
Environmental degradation is causing alterations of the landscapes where most 
species live. For migratory species that are composed of populations occupying 
ranges that span continents, deciphering when and where alterations are most 
impacting species requires considering their full annual cycle ecology (Wilcove and 
Wikelski 2008, Marra et al. 2015). Here, we found different habitat preferences 
across periods of the annual cycle, suggesting wood thrush populations will respond 
differently to changes in habitat composition and configuration during different 
periods of the annual cycle. For example, relaxation of habitat preferences during 
migration may be an adaptive trait to facilitate a fast migration, but across rapidly 
changing landscapes, it could lead to birds occupying unsuitable stopover habitats. 
Recent work has found high use of human dominated landscapes by migrating 
songbirds, but whether these represent suitable habitat or ecological traps is not 
known (La Sorte et al. 2014, Zuckerberg et al. 2016). Developing a better 
understanding of the vulnerability of migrating wood thrush to landscape changes 
32 
 
 
 
may be particularly important at spring Gulf of Mexico stopover sites where the 
entire population of wood thrush appear to spend time during spring migration. 
Unfortunately, our knowledge of how species respond to environmental heterogeneity 
in the non-breeding season is lacking, even for well-studied species such as wood 
thrush (Faaborg et al. 2010). As technologies improve, the increased temporal and 
spatial resolution of GPS tags will allow researchers to investigate not only how 
ecological responses of populations and individuals change across the annual cycle 
but also the factors driving these variations within and across populations and 
individuals.  
Acknowledgements 
Funding for this study came from Strategic Environmental Research & Development 
Program (RC-2121), Minnesota National Park Service, Bedford Audubon Society, 
Forsyth Audubon Society, International Audubon Society and a United States 
Department of Agriculture McIntire-Stennis Capacity Grant. We thank numerous 
field assistants at our field site in Indiana, Tim Guida and Peter Keller, as well as the 
volunteers from Forsyth Audubon, Bedford Audubon, Carpenter Natura Center, and 
Warner Nature Center. In particular, we would like to thank Kimberly Brand, 
Zachary Landin, Janelle Hope Robbins, Paul Smithson, Katherine Thorington and 
Jennifer Veith for their logistical support in the field. We would also like to thank 
Jonathon J. Valente for assistance with Bayesian models used in the analysis. 
33 
 
 
  
Tables 
Table 1.1 Repeatability estimates of habitat features used by wood thrush tracked 
with GPS tags, calculated between stationary (breeding and winter) and migratory 
periods (fall and spring migration). 
 Repeatability 
 Stationary Migratory 
Habitat Features  n R p-value n R p-value 
Patch-level       
Tree Cover (%) 181/186 0.327 <0.0001 92/51 0.043 0.051 
EVI 173/202 0.128 <0.0001 85/48 0.051 0.274 
Patch Size (km2) 184/214 0.269 <0.0001 92/59 0.072 0.076 
Proximity to Non-Forest (m) 184/163 0.699 <0.0001 91/55 0.047 0.28 
Landscape-level       
Tree Cover 1km (%) 184/212 0.380 <0.0001 92/52 0.022 0.406 
Average Patch Size (km2) 184/212 0.379 <0.0001 92/52 0.032 0.329 
Notes: Samples sizes (n) are given during stationary period: breeding/winter; migratory period: fall 
migration/spring migration. Significant effects in boldface. 
 
 
 
 
 
 
 
 
 
 
 
 
 
34 
 
 
  
Table 1.2 Estimated slope of the seasonal trend based on a generalized linear model 
examining the effects of the interaction between breeding longitude and local and 
landscape scale habitat features of habitats occupied by wood thrush from 5 breeding 
populations across the annual cycle. Estimated season-specific trends and 95% 
confidence intervals. 
Model Season Estimated Slope 95% CI 
Patch Level    
Tree Cover (%) n/a   
    
EVI n/a   
    
Patch Size (ha) Breeding -344.0 3.7, 650.9 
 Fall Migration -116.4 -218.1, 450.8 
 Winter -276.6 -2.3, 576.1 
 Spring Migration 0.3 -381.2, 387.1 
    
Prox. to Non-Forest (m) Breeding -0.082 -0.2, -0.009 
 Fall Migration 0.083 0.004, 0.2 
 Winter -0.18 -0.3, -0.1 
 Spring Migration 0.01 -0.082, 0.10 
    
Landscape level    
Avg. Tree Cover 1 km (%) Breeding -0.47 -0.98, 0.10 
 Fall Migration 0.0079 -0.57, 0.59 
 Winter 0.41 -0.12, 0.95 
 Spring Migration 0.20 -0.46, 0.87 
    
Avg. Patch Size 1 km (ha) Breeding -68.4 -364.3, 227.3 
 Fall Migration 245.6 -75.0, 566.2 
 Winter -313.4 -603.3, -24.2 
 Spring Migration 103.6 -276.7, 483.9 
    
Land Cover 5km n/a   
Notes: Values in bold have confidence intervals that do not include zero. 
 
35 
 
 
  
Figures  
Figures Legend 
Figure 1.1. Conceptual hierarchy of the decision-making process of habitat use by a 
migratory songbird. At higher spatial scales (A) and (B) the process is expected to be 
largely inflexible and constrained by extrinsic habitat factors. At lower spatial scales 
(C) and (D) the process likely involves the assessment of the intrinsic factors of the 
habitat based on cues and exploratory assessment. Different patterns of migratory 
connectivity are illustrated (A). Strong connectivity (solid lines) occurs when most 
individuals from one breeding population move to the same nonbreeding locations. 
Weak connectivity (dashed lines) occurs when individuals from the same breeding 
population move to different nonbreeding locations.    
 
Figure 1.2. Migratory connectivity of a migratory songbird, the wood thrush, obtained 
from GPS-tags deployed across 5 breeding populations in the United States. Across 
maps, colours correspond to breeding location (red = DE, blue = IN, green = MN, 
purple = NY, orange = NC). (A) Locations (breeding, stopover and winter) and 
migratory routes of wood thrush (n = 19).  Lines connect consecutive location fixes 
during fall (solid) and spring (dashed) migratory period and do not represent direct 
path travelled. (B) Stopover locations at the northern coast of the Gulf of Mexico 
during fall (solid circle, n = 14) and spring (open sqaure, n = 12) migration. (C) 
Winter locations of wood thrush, shapes denote different individuals. Eight wood 
thrush occupied two or more winter locations. 
36 
 
 
  
Figure 1.3. Seasonal changes in average percent of locations found in each of five 
land cover class for wood thrush tracked using GPS geolocators from 5 breeding 
populations across the annual cycle. Land cover classes derived from 5-km buffer 
around location estimate and include agriculture, mixed agriculture and natural 
vegetation, forest, wetlands and wooded savannah. For each breeding population, 
seasonal sample sizes are indicated in the brackets. 
 
Figure 1.4. Parameter estimates and associated 95% credible intervals for the top 
discrete choice model examining habitat selection patterns of individually tagged 
wood thrush at multiple spatial scale during a) breeding, b) winter, c) fall migration 
and d) spring migration. For each period discrete choice models were run at multiple 
extents: 0.5 km (closed square), 5 km (closed circle), 15 km (open square) and 25 km 
(closed square). Local habitat features include percent tree cover, Enhanced 
Vegetation Index (EVI), patch size (km2) and proximity to non-forest. Landscape 
habitat features include average percent tree cover and average patch size (km2). Not 
all habitat features were present in all top-models.  
37 
 
 
  
Figure 1.1 
 
 
 
38 
 
 
  
Figure 1.2 
 
 
39 
 
 
  
Figure 1.3 
 
40 
 
 
  
Figure 1.4 
 
41 
 
 
  
 : Patterns of moisture and food availability during the 
non-breeding season influence small- and large-scale space use 
strategies in a migratory songbird 
 
Abstract 
Understanding the ecological correlates driving space use strategies in species can be 
critical for predicting population dynamics; however, such information can be 
difficult to attain for small mobile species such as migratory songbirds. We combined 
radio telemetry and high-resolution GPS tracking to examine small- and large-scale 
space use strategies under different moisture gradients for wood thrush (Hylocichla 
mustelina), a Neotropical migratory songbird. Food availability is often thought to be 
a key ecological factor dictating space-use patterns in animals and in this study, we 
explored the role moisture plays in driving food abundance and in turn, space-use 
strategies at a non-breeding site in Belize across three years. Overall, patterns of 
moisture on non-breeding home ranges influences food availability and both small- 
and large-scale movement patterns. Individuals with home ranges in drier habitats 
experienced lower food abundance, reduced body condition and occupied larger 
home ranges. Contrary to expectations given previous research, with two types of 
tracking technologies we discovered wood thrush exhibit low rates of seasonal site 
persistence across their non-breeding range and that individuals in wetter habitats 
were more likely to engage in facultative and permanent mid-winter shifts up to 148 
km between home ranges. We suggest that these movements may be a condition 
dependent strategy to enable non-breeding wood thrush to find alternative habitat as 
food availability declines at the end of the dry season. Given that increased aridity is 
42 
 
 
  
predicted across the non-breeding range of wood thrush, future research should delve 
deeper into understanding how moisture availability impacts within and between 
season dynamics and ultimately the population dynamics of this declining species.   
Introduction  
Addressing recent declines in long-distant migratory birds requires an understanding 
of the processes limiting populations across the annual cycle (Marra et al. 2015). One 
major gap is the role of ecological factors (e.g., rainfall, predation, food and their 
interactions) in driving population dynamics while birds are on their tropical 
nonbreeding ranges (Faaborg et al. 2010). To gain a better understanding of the role 
of ecological factors it is first necessary to determine the space-use strategies of 
migrants during the nonbreeding period.  It has often been assumed that most species 
of migratory birds spend their non-breeding season on a single home-range.  
Evidence is accumulating, however, that suggests they exhibit a broad range of space-
use strategies (Heckscher et al. 2015, Kole?ek et al. 2018). Aside from a few early 
studies with marked (Moreau 1972, Morton 1980) or radio-telemetered individuals 
(Rappole et al. 1989), it was not until the advent of modern tracking technologies that 
researchers have begun to recognize the extent of inter- and intra-specific variation in 
space-use strategies on the non-breeding range of migratory birds (McKinnon et al. 
2013). From facultative short-distance shifts in home ranges (Smith et al. 2011b, 
Schlaich et al. 2016), to obligate long-distance intra-tropical migrations (Renfrew et 
al. 2013, Thorup et al. 2017), tracking studies are uncovering a wide range of non-
breeding movement behavior from ecologically and socially distinct species 
(Stutchbury et al. 2016).  
43 
 
 
  
Elucidating the ecological factors underlying the diversity of movement 
strategies has been more limited. Food availability has been identified as a key 
ecological factor contributing to non-breeding season limitation (Sherry et al. 2005, 
Brown and Sherry 2006b, Diggs et al. 2011) and winter space use strategies (Zahavi 
1971, Brown and Sherry 2008, Smith et al. 2011b). In wintering American Redstarts 
(Setophaga ruticilla), experimental food reductions on individual territories led some 
individuals to vacate territories and adopt a non-territorial strategy (Cooper et al. 
2015). Furthermore, these individuals experienced a reduction in pectoral muscle 
mass and, on average, a one-week departure delay on spring migration, suggesting 
food limitation impacted physical condition and ultimately influenced space-use 
decisions.  At larger spatial scales, tracking studies of Palearctic migrants have begun 
to show that intra-tropical migrations on non-breeding ranges in sub-Saharan Africa 
follow shifts in proxies of food abundance (e.g. vegetation indices; Trierweiler et al. 
2013, Thorup et al. 2017). Therefore, food availability is likely a key factor driving 
alternative space use strategies; however, the extent to which it dictates space-use 
dynamics among individuals, populations and species remains poorly resolved.   
In much of the tropics differential food availability across habitat types has 
been shown to be driven largely by gradients in moisture, primarily through its effect 
on phytophagous insects (Janzen 1973, Wolda 1978, Studds and Marra 2011a). For 
migratory songbirds in the tropics, this is of particular importance as aridity increases 
towards the end of the dry season (due to seasonal shifts in rainfall), at the same time 
birds are preparing for spring migration. For small migratory warblers, non-breeding 
rainfall can have important consequence for both spring migration phenology and 
44 
 
 
  
annual survival (McKellar et al. 2013a, Rockwell et al. 2017). Therefore, how 
migratory birds handle increasingly arid conditions as the non-breeding season 
progresses and the role this plays in shaping space-use strategies and movement 
dynamics could have important implications for our understanding of non-breeding 
ground limitations for migratory birds.   
The aim of this study was to explore the ecological correlates of movement 
behaviour for wood thrush (Hylocichla mustelina), a long-distance Neotropical 
migratory songbird.  We tested the hypothesis that moisture indirectly drives food 
abundance, and this corresponds to space use strategies in wood thrush. Specifically, 
we tested the prediction that wood thrush in wetter relative to drier sites will (1) 
occupy smaller home ranges, (2) have access to more fruit and insect resources, (3) 
maintain better energetic condition, and (4) have higher seasonal site persistence. 
Using radiotelemetry, we quantified local home ranges for wood thrush as well as 
moisture levels and food availability on these corresponding areas for individual birds 
at a non-breeding site in Belize. To explore the influence of patterns of moisture on 
seasonal site persistence and movement dynamics at larger spatial scales we paired 
high-resolution GPS tracking data, deployed on 5 breeding populations, to satellite 
derived indices of wetness. By combining results from field data and remote tracking, 
we were able to examine the influence of moisture patterns on food availability at 
both small- and large-spatial scales.  
 
45 
 
 
  
Methods 
Study Species  
Wood thrush are a large (40-50g), omnivorous Neotropical migratory songbird that 
breed in the forests of eastern North America and winters primarily in the lowland (up 
to 500m) tropical forests of Mexico and Central America. Although a common forest 
songbird, their global population size has declined by 60% since 1966 based on 
breeding surveys (Sauer et al. 2013). During the non-breeding season wood thrush are 
solitary and radio-tracking studies have documented two non-breeding space-use 
strategies. Some individuals maintain flexible territories, while others 
opportunistically float across territories, a strategy that has been associated with 
higher mortality (Rappole et al. 1989).  
Field Methods: 
Research was conducted at two adjacent lowland tropical evergreen forest reserves in 
the late dry season (mid-January through mid-April) from 2013 to 2015. Both forest 
reserves were located along the Bladen river at the base of the Maya Mountains in the 
Toledo District of Belize. The Belize Foundation for Research and Environmental 
Education consists of 467 ha of secondary forest and early successional scrub habitat 
due to heavy damage sustained during Hurricane Iris in 2001. The Bladen Nature 
Reserve consists of 39,270 ha of protected rainforest managed by the Ya?axch? 
Conservation Trust and was less damaged by Hurricane Iris resulting in a mix of 
secondary and old growth forest in our study area. The area receives an average 
46 
 
 
  
rainfall of 2500 mm/year and experiences a pronounced dry season that lasts from 
February-April.  
Birds were captured in mist nets using conspecific playback during two 
sampling periods in order to monitor individuals across the dry season. The first 
sampling period ran from mid-January until mid-February hereafter ?early?. We 
attempted to recapture individuals during a second sampling period in mid-March, 
hereafter ?late?.  At initial capture, all individuals were marked with a U.S. 
Geological Service aluminum band and unique color-band combination for 
identification. Birds were aged by feather characteristics (yearling or adult).  
Individuals were weighed to the nearest 0.1 g, and morphological measurements 
taken to estimate body size (unflattened wing chord, tarsus, retrix length and bill 
length). Visible fat deposits were scored following (Holmes et al. 1989) and pectoral 
muscles were palpated and scored following Gosler (1991). Blood samples (100-150 
?L) were collected from the brachial vein and stored in Queen?s lysis buffer for 
genetic sexing (detailed methods in Appendix S4). Blood and claw samples were also 
collected for stable isotope analysis (detailed methods in Appendix S4).  
Radio-telemetry  
To define home-ranges and monitor seasonal site persistence a total of 168 birds were 
equipped with either a VHS radio transmitters (n = 17, model BD-2, Holohil Systems 
Ltd., Carp, Ontario, Canada; n = 119, model PipAg393, Lotek Wireless, Newmarket, 
Ontario, Canada) or coded radio-transmitter (n = 45, model NTQB-4-2, Lotek 
Wireless). Transmitters were attached using a leg-loop harness constructed of Stretch 
Magic and secured by crimp beads (Rappole and Tipton 1991).  We radio-tracked a 
47 
 
 
  
total of 98 adults (n = 38 females, n = 50 males, n = 10 unknown) and 78 yearlings (n 
= 32 females, n = 37 males, n = 9 unknown). Thirteen individuals were tracked in 
multiple years, and unless otherwise indicated, only their first year of data was 
included in analyses. Individuals with VHF transmitters were located by homing 
using a Biotracker receiver (Lotek Wireless) and three-pronged flexible Yagi antenna 
(Lotek Wireless), individuals with coded tags were located using an SRX receiver 
(Lotek Wireless). Upon locating a bird, the position was taken with a handheld global 
positioning system unit (Garmin GPSMap). Location data was collected a maximum 
of twice per day during either morning and/or evening tracking sessions. 
When an individual failed to be detected a systematic search was initiated by 
walking up to 1 km (minimum 500-m) in all 4 cardinal directions. If the search failed, 
we would repeat the procedure in the following days and periodically throughout the 
season (approximately bi-monthly). Individuals were tracked until either (1) the bird 
departed the site or (2) the transmitter was recovered. Individuals whose transmitters 
were recovered (e.g. due to mortality, harness failure) were excluded from analyses of 
seasonal site persistence. Some cases of tag failure or mortality are likely to be 
included as site departures, however extensive recapture efforts suggest tag failures 
were rare.  
Home-range area was defined as the 95% isopleth of the utilization 
distribution estimated from an area-corrected kernel density estimator which reduces 
bias due to small sample sizes (Fleming and Calabrese 2017). We used the ctmm 
package in R (Calabrese et al. 2016) to derive utilization distributions (detailed 
48 
 
 
  
methods in Appendix S4). We restricted analysis to individuals with 16 or more 
relocations across the season (n = 53), inspection of confidence intervals indicated 
area estimates stabilized around 16 locations.    
Body Condition 
To examine the influence of moisture on body condition we (1) examined the initial 
condition of individuals in late January and early February and (2) the change in 
condition of individuals that were recaptured in mid-March. Initial body condition 
was assessed using a body condition index derived by predicting the lean body mass 
of each bird based on a regression equation of mass versus unflattened wing chord, 
tail and tarsus for fat-free birds with a pectoral score of 1 (Strong and Sherry 2000, 
McKinnon et al. 2015a). Body mass was then compared to the predicted lean body 
mass based on structural size, and the percent difference was calculated (Predicted 
Mass ? Measured Mass / Predicted Mass). For birds that were successfully recaptured 
in mid-March we examined change in condition by directly measuring change in 
mass between the early and late sampling period.  
Diet sampling 
The influence of moisture on food availability and diet was assessed using two 
methods: (1) environmental fruit and insect sampling and (2) isotopic analysis. To 
characterize food availability, fruit and insect surveys were performed at tracking 
locations selected randomly without replacement for each individual. In 2013, insect 
and fruit surveys were conducted at 3 and 1 randomly selected tracking locations, 
respectively. Following 2013 both insect and fruit surveys were performed at 5 
49 
 
 
  
randomly selected tracking locations to provide better coverage of wood thrush home 
ranges. Surveys were performed twice to assess change in food availability across the 
dry season. Early sampling period surveys were conducted in mid-February on all 
individuals with a minimum of 8 relocations, late sampling period surveys were 
conducted in late-March for all individuals that remained on site until the second 
week of March.  Fruit surveys were performed by counting all fleshy fruit in a 5-m 
radius (ground to canopy). Insect surveys were performed by placing a 0.5 m2 frame 
on the ground and performing a 5 min search to identify all arthropods to order and 
size class (head to abdomen; Strong and Sherry 2000). Due to low site-persistence we 
focused analysis on surveys obtained for individuals during the first sampling period 
(n = 44 fruit surveys; n = 43 insect surveys). 
To quantify how the proportion of insects in the diet of wood thrush was 
influenced by moisture we used stable nitrogen isotopes (?15N) derived from blood 
and claws as an index of food consumption (versus available food resources; detailed 
methods in Appendix S4). In general, tissues become more enriched in heavier stable-
nitrogen isotopes as their diet includes protein from higher tropic level food items 
(Hobson 1993, Pearson et al. 2003). Different turnover rates of tissues allow the 
integration of dietary information over multiple time scales (Bearhop et al. 2002, 
Bauchinger and McWilliams 2009). Due to the longer turnover rate of claws (half-life 
= 35 days) relative to whole blood (half-life = 11-16 days) we expect claws to provide 
an indication of diet integration over a few months, while whole blood provides an 
integration of diet over several weeks (Pearson et al. 2003, Louren?o et al. 2015).  
50 
 
 
  
Patterns of Moisture 
Patterns of moisture were quantified using (1) an index of moisture derived from 
remote sensing imagery and (2) stable-carbon isotopes. To derive an index of 
moisture from satellite imagery we calculated the wetness metric from the tasseled-
cap transformation of Landsat images (Crist and Kauth 1986). This metric has been 
extensively used in tropical landscapes and reflects soil and canopy moisture (Cohen 
and Goward 2014). We performed the transformation on bi-monthly 30-m resolution 
Landsat images obtained from our study area for the period of December to April 
2012 to 2015 (detailed methods in Appendix S4). For each image, we extracted the 
mean wetness metric for each (i) home range and (ii) plot. At each home range and 
plot we calculated the average and coefficient of variation (CV) of wetness (1) across 
the dry season and (2) separately across the early (Dec ? Feb) and late (Mar-Apr) dry 
season.  
Additionally, we used stable-carbon isotopes from blood and claws as an 
index of water availability within the habitats utilized by wood thrush (?13C; detailed 
methods in Appendix S4). Differences in stable-carbon isotopes are largely driven by 
photosynthetic pathways in plants, with C3 plants being more depleted than C4 
plants, such as grasses. However, within C3 plants, which dominate forested habitats, 
water-use efficiency also influences ?13C values, with C3 plants becoming more 
enriched as water availability decreases (Michener and Lajtha 2007, Kohn 2010). 
Similar to results found in other studies we found that ?13C values correlated well 
with satellite derived wetness indices (Appendix S5). As with stable-nitrogen isotopes 
51 
 
 
  
(above), difference in turnover rates between claw and blood tissue allows us to 
examine changes in environmental conditions over different time windows.  
GPS Tracking 
To assess seasonal site persistence at a larger spatial scale we examined over-winter 
location data obtained from archival GPS transmitters (Model PinPoint-50, 1.8 g, 50 
fixes, Lotek Wireless) deployed on adult male wood thrush from 5 breeding 
populations (n = 137; Delaware, Indiana, North Carolina, New York and Minnesota; 
Appendix S1 Table S1.1).  The GPS transmitters provide location estimates every 2 
weeks during the winter period for individual birds with an estimated accuracy of 10 
meters. We obtained location estimates from 18 individuals across the winter period, 
with an average of 11 ? 3.9 location estimates per individual (? SD; range 2-17). We 
defined a dispersal event as any movement greater than 1 km after a stationary period 
(based on 2 or more locations in the same area) that occurred prior to the 
commencement of the spring migratory period. Distance between successive points 
(step length) was calculated using the move package in R (Kranstauber et al. 2018). 
The commencement of the spring migratory period was defined by the departure of 
the first tracked individual on migration (identified as consistent northward 
movements over consecutive days). To quantify how moisture patterns influenced 
seasonal site persistence of individuals equipped with GPS transmitters we derived 
the wetness metric (see above) from tasseled-cap transformed Landsat images 
interpolated for each location estimate (detailed methods in Appendix S4). 
52 
 
 
  
Data Analysis 
To determine the influence of moisture patterns on home range size we constructed 
generalized mixed effect models. We generated a set of candidate models that 
reflected a priori hypotheses for home range size. We hypothesized that home range 
size changes with food availability (fruit and insect surveys, ?15N). We also included 
moisture directly in the model (wetness, ?13C) to account for other potential 
mechanisms that may influence home range size (e.g. predation). Additionally, we 
hypothesized that condition (at capture) and dominance hierarchies between sex and 
age classes would affect home range size. Year was included as a random effect. For 
both ?13C and ?15N, blood and claw stable-isotope values were highly correlated; 
therefore, for all response variables separate model sets were constructed including 
stable-isotope values derived from either blood or claw.  Average wetness and the 
coefficient of variation of wetness was highly correlated in our study site in Belize; 
therefore, only average wetness was included in model sets.  
We examined the influence of moisture levels on food availability and diet by 
constructing separate mixed effect models with fruit counts, insect counts and ?15N 
as the response variable. We developed a candidate model set for each response 
variable based on the predictions that moisture, measured as home-range wetness, 
will directly influence food availability. We also hypothesized that condition at 
capture, home range size and dominance relationships between age and sex classes 
may influence food availability and diet. Fruit and insect counts were log-transformed 
and fit with a general linear model. For the analysis of ?15N, linear mixed models 
were constructed. Sampling period was included as a fixed effect and period specific 
53 
 
 
  
estimates of plot wetness were used. Bird identity was nested within year as a random 
effect.  
  To determine the influence of moisture patterns on body condition index 
during the early sampling period we examined separate linear mixed effect models for 
isotopes derived from blood and claws. We hypothesized that moisture would 
indirectly influence body condition through its effect on food availability as measured 
by stable-nitrogen values. We also included measures of habitat moisture, a satellite-
derived wetness metric and stable-carbon values to account for other possible 
mechanisms in which moisture may influence body condition. We also included sex 
and age classes to explore the influence of dominance relationships, and year was 
included as a random effect. For individuals recaptured in the second sampling period 
we also determined the influence of moisture patterns on weight change between 
sampling periods. We evaluated the same set of hypotheses as proposed for body 
condition during the early sampling period.  For individuals that remained in our 
study area until the end of the dry season we also determined how conditions and 
food availability changed over time. We used a paired-t test to compare mass across 
sampling periods for recaptured individuals. To assess change in food availability we 
performed separate Wilcoxon signed-rank tests to compare fruit and insect counts on 
home ranges.  
We modeled seasonal site persistence (time-to-event outcome) using the Cox 
proportional-hazards regression model (CPH; Cox 1972, Therneau and Grambsch 
2000). CPH models allow estimation of the effect of covariates on the hazard of 
departure (hazard ratio), or risk of departure. CPH models were performed using the 
54 
 
 
  
survival package in R (Therneau 2015). We developed a set of a priori models to 
examine the influence of moisture on seasonal site persistence for radio-tracked birds 
in Belize. Similar to above, we examined the indirect impact of moisture on food 
availability as measure by ?15N derived from claws. We also included direct 
measures of moisture, satellite-derived plot wetness and ?13C derived from claws, to 
account for other mechanisms driven by moisture levels. We also included condition 
at capture, age and sex class to explore the influence of dominance hierarchies (Marra 
2000). Software to perform random effects CPH models are currently limited and do 
not provide methods to test model assumptions; therefore, year was included as a 
fixed effect. For individuals tracked using GPS tags we constructed a more limited 
candidate set that included wetness, the CV of wetness and winter latitude. For 
individuals that performed intra-winter movements we also compared wetness and the 
CV of wetness between locations pre- and post-movements. We tested if wetness or 
the CV of wetness differed using a paired t-test.  
All candidate model sets were ranked using an information theoretic approach 
by comparing Akaike?s information criteria corrected for small sample size using the 
R package MUMIn (Burnham and Anderson 2002, Barton 2017). All models with 
?AICc  ? 2 were considered equally plausible. When more than one model was 
supported, we performed unconditional model averaging to derive coefficient 
estimates and 95% confidence intervals.  Although models ?AICc up to 4 are also 
considered well supported (Symonds and Moussalli 2011), the inclusion of more 
competing models did not yield biologically relevant differences in the final models 
(based on inspection of confidence intervals) and therefore were not included. All 
55 
 
 
  
analyses were conducted in R 3.4.3 (R Core Team 2017). Generalized mixed effect 
models were run in the R package lme4 (Bates et al. 2015).  Linear mixed effects 
models were run in the R package nlme (Pinheiro et al. 2017) with restricted 
maximum likelihood estimation. General linear models were run in the R package 
stats (R Core Team 2017). All models were assessed for violation of model 
assumptions by visual inspection of plots of residual versus predicted values and 
quantile-quantile plots. CPH models were checked for proportional hazards using the 
cox.zph() function in the survival package. 
Results 
The majority of radio-tracked wood thrush (68%; n = 115) departed our study sites in 
southern Belize prior to the start of the spring migratory period (Julian date range: 25 
? 116; Appendix S4 Table S4.1). Despite extensive efforts to relocate individuals, we 
were unable to locate new home ranges for departed individuals successfully. As the 
dry season progressed, there were significant changes in environmental conditions on 
our study site. In general, home-range wetness (n = 75, z = -2.01, p = 0.045) and 
insect abundance declined (n = 75, z = -2.51, p = 0.013) while fruit abundance 
increased between our two sampling periods (n = 78, z = -3.30, p < 0.001; Appendix 
S4, Table S4.1, Figure S4.1). Home range wetness was significantly lower in the 
2013 season compared with 2014-2015 (F2,27 =  1270, p <0.001, n = 52; Bonferroni 
adjustment: 2013-2014, p <0.001; 2013-2015, p <0.001; 2014-2015, p = 1; Table 
S4.1) and corresponded to overall lower dry season rainfall across Belize (The World 
Bank: Climate Change Knowledge Portal: Climatic Research Unit of University of 
East Anglia).  
56 
 
 
  
Home-range Size 
For all birds that remained site-persistent until approximately mid-March we obtained 
sufficient relocations (>16 relocations) to calculate 95% kernel density home-range 
estimates. Most individuals (n = 44) occupied a contiguous forested area moving an 
average of 55.8 ? 89.20 m between relocations. However, a number of individuals (n 
= 9) made regular off-territory forays (up to 2 km away) resulting in less contiguous, 
larger home ranges. Average home-range size across the three years of this study was 
2.3 ? 2.66 ha (?SD, median = 1.35 ha, n = 53; Table S4.1). We found evidence that 
moisture, as measured by ?13C in blood, influenced home range size (Figure 2.1, 
Appendix S4 Table S4.2). Individuals with more enriched ?13C (in drier habitats) in 
their blood, but not claw, occupied larger home ranges (Table 2.1). The model 
including claw isotopes, which represents diets incorporated over a few months, did 
not find an effect of ?13C on home range size, but instead found that female and 
yearling birds were more likely to occupy larger home ranges (Figure 2.1, Table 2.1, 
Appendix S4 Table S4.1 and Table S4.2).  
Food availability and diet 
Home range wetness was included in the top model explaining both fruit counts and 
insect abundance (Figure 2.1; Appendix S4 Table S4.1 and Table S4.3). Individuals 
with home ranges that were wetter had more fruit (? = 15.66, 95% CI = 9.97, 21.35) 
and insects (? = 12.83, 95% CI = 6.54, 19.13). We also quantified food consumption 
using stable nitrogen isotopes to assess the proportion of insects versus fruit in 
individuals? diets. We found evidence that plot wetness and sampling period 
influenced the proportion of insects consumed in the diet (Figure 2.1, Appendix S4 
57 
 
 
  
Table S4.4). Claw ?15N values, which represent stable isotopes incorporated at the 
timescale of months, were more enriched (higher insect consumption) when study 
plots were drier (? = -4.66, 95% CI = -8.75, -0.57). Although wetness was included in 
the top model describing blood ?15N values, which represents stable isotopes 
incorporated at a weekly time scale, the confidence intervals included zero. The best 
predictor of blood ?15N values was sampling period, individuals had more depleted 
?15N values (fewer insects consumed) during the late sampling period (? = -0.46, 
95% CI = -0.72, -0.20). 
Body Condition 
There was a decline in the body condition of wood thrush across the late dry season, 
with individuals losing on average 1.4 ? 2.7g (? SD; percent body mass: 2.9 ? 5.83% 
SD; < 2 months; t = 3.23, p = 0.0013). Based on the averaged model including stable 
isotopes derived from blood, we found evidence of an effect of moisture, diet and an 
interaction between age and sex class on body condition early in the season (Figure 
2.2, Appendix S4 Table S4.1 and Table S4.5). Individuals in wet habitats, as 
measured by ?13C values in blood, were in better condition (Table 2.2). Body 
condition was also higher in individuals with enriched ?15N values, indicative of 
higher insect consumption. Finally, the averaged model predicted body condition 
would be highest in adult males while lowest in adult females (Figure 2.2a). Similar 
effects were found in the model that examined claw isotope values (Appendix S4 
Table S4.5). Given that blood isotopes have a shorter turnover rate that would better 
reflect habitats used prior to capture and current body condition we focus on results 
derived from blood stable isotopes.  Change in body mass was negatively related to 
58 
 
 
  
?13C signatures in blood but not claw (Figure 2.2d; Appendix S4 Table S4.6). 
Individuals with depleted ?13C values in blood, characteristic of drier habitats, lost 
more weight over the sampling period (Table 2.3). For the model that included stable-
carbon derived from claw, the confidence intervals for each parameter estimates 
overlap with zero (Table 2.3). 
Seasonal Site Persistence 
At the beginning of the spring migratory period (late April), the daily probability of 
seasonal site persistence was 20.0% (95% CI 0.13 ? 0.31; n = 23/92 birds remained 
on site). We found evidence of an effect of habitat moisture on the probability of site 
persistence (Figure 2.3; Appendix S4 Table S4.7). Individuals with depleted ?13C in 
claws, indicative of wetter habitats, had a higher departure probability (? = -1.28, 
95% CI = -2.04, -0.51).  The probability of departure was also higher in 2015, a 
relatively wet year (? = 1.43, 95% CI = 0.53, 2.33).  
Seasonal site persistence was assessed at a larger spatial scale by examining 
the winter locations of wood thrush equipped with GPS archival pinpoint tags (Figure 
2.4). Overall seasonal site persistence was higher in our GPS tracked sample. At the 
commencement of the spring migratory period probability of site persistence was 
66% (95% CI = 0.44 ? 0.97; n = 7/17 individuals) compared with 20.0% (95% CI 
0.13 ? 0.31) from birds tracked with radio-transmitters from our study site in Belize. 
The first GPS tracked individual departed its initial home range at the end of February 
(Appendix S4 Table S4.1). Similar to wood thrush in Belize, site persistence was 
influenced by both wetness and variability in wetness (Figure 2.3; Appendix S4 Table 
S4.8), and probability of departure was higher for individuals in wetter habitats (? = 
59 
 
 
  
83.52, 95% CI == 12.15, 154.89). In addition, probability of departure was higher for 
individuals occupying areas with higher variation in wetness (? = 147.33, 95% CI = 
8.12, 286.53). On average, individuals that departed initial home ranges moved 48.2 ? 
57.8 km (range: 1.3 ? 182.3 km; Figure 2.4); whereas, individuals that remained site 
persistent moved an average of 124 ? 159.5 m (?SD) between relocations (every 2 
weeks). There was no difference in the average wetness or variability in wetness of 
locations occupied by individuals pre- or post-movement (average wetness: t = 0.68, 
p = 0.263; CV wetness: t = 1.07, p = 0.168). 
 
Discussion 
By combining on the ground field work at a single site with high-resolution tracking 
data from GPS transmitters we were able to demonstrate that both fine-scale space 
use strategies and larger scale movement dynamics of non-breeding wood thrush 
sampled from throughout their breeding range were dependent on patterns of 
moisture which influenced food availability. At our winter study site in Belize, 
individuals in drier habitats had lower food resources, were in poorer physical 
condition and occupied larger home ranges. We also showed that as habitats became 
more arid across the dry season, insect abundance declined (although fruit abundance 
increased) and birds in drier habitat lost weight. Overwinter site persistence was low 
(0.20, 95% CI 0.13 ? 0.31), and contrary to expectations, departure probability was 
higher for birds occupying wetter habitats with higher food availability. GPS tracking 
confirmed this pattern (regardless of breeding origin) and revealed that individuals 
can make permanent mid-winter shifts of home ranges and move large distances to 
60 
 
 
  
settle in new areas (range: 1.3 ? 148 km). These results provide evidence that space-
use strategies in overwintering wood thrush are flexible, and at least partly driven by 
fluctuating environmental conditions linked to patterns of moisture.  
Mounting evidence points to the importance of food availability as a key non-
breeding resource for migratory songbirds and the major driver of individual space-
use behavior (Greenberg and Salewski 2005, Sherry et al. 2005, Smith et al. 2011a, 
McKinnon et al. 2015a). The results presented here support the conclusion that 
habitat quality is linked to food availability, which is ultimately driven by patterns of 
moisture. In drier habitats (based on stable-carbon isotope values) with lower fruit 
and insect abundance, individuals were more likely to occupy larger home-ranges, 
often using multiple, non-contiguous areas. Previous studies suggest that birds will 
adopt alternative space use strategies when optimal habitats are saturated to maximize 
survival under suboptimal conditions or when food resources are unpredictable 
(Zahavi 1971, Brown and Sherry 2008). Insects are a relatively stable food resource 
whereas fruit is extinguished more quickly and does not replenish within a season. 
Wood thrush were often observed making forays (up to 2 km) to large fruiting trees 
(e.g. Trophis racemose), which could explain the increase in home range size for 
some individuals. However, we found no decrease in the proportion of insects 
consumed by individuals on larger territories based on stable nitrogen-isotope levels 
of tissues. Instead, we found that wood thrush consumed larger proportions of insects 
when plots were drier.  Therefore, we suggest that larger home ranges may be an 
adaptive strategy to compensate for lower insect availability.  
61 
 
 
  
Evidence for differential space use by subordinate age and sex classes through 
behavioral dominance has also been increasing (Marra 2000, Diggs et al. 2011, Catry 
et al. 2012).  Consistent with this, we also found evidence that social dominance 
influenced space-use strategies in this system. Both females and young birds were 
more likely to inhabit larger home ranges and had lower body condition suggesting 
more aggressive adult males may exclude females and young birds from limited high-
quality sites. Together, these differences in space use behavior suggest that there is 
competition for high quality habitats on the non-breeding grounds of wood thrush and 
less competitive birds are relegated to sub-optimal, drier, habitats where larger home 
ranges might be required to cope with lower food availability. In American redstarts, 
sexual habitat segregation due to competitive exclusion by dominant adult males 
forces subordinate individuals into drier habitats with lower food abundance (Marra 
2000). Furthermore, occupancy of suboptimal scrub habitat has been found to carry-
over to the breeding season resulting in reproductive consequences for excluded 
individuals of both sexes (Norris et al. 2004, Reudink et al. 2009). However, the 
effect of age and sex on home range size was not mirrored in our analysis of stables 
isotopes derived from blood, which provides information on environmental 
conditions across more recent time scales. This may indicate that as conditions 
become drier across plots, which would be better captured by isotopes derived from 
blood, moisture gradients may become a better predictor of home range size, 
regardless of age or sex. Given that individuals in drier habitats were in poorer 
condition and were more likely to lose weight across the dry season, the adoption of a 
larger home range may not fully compensate for lower food availability under all 
62 
 
 
  
conditions. Adopting a larger home range, may therefore have consequences for 
individual performance that could be exacerbated at the end of the dry season when 
conditions are driest, and individuals need to prepare to migrate back to their breeding 
grounds.  
The permanent mid-winter shifts of home ranges we discovered, in both our 
study population in Belize and among an independent group of GPS tracked birds 
from throughout the breeding range of the wood thrush, was unexpected. The fact that 
we found similar relationships between ecological correlates and space-use behavior 
with both approaches, however, provides us with confidence in the biological 
importance of these findings. Prior research on this species have shown that wood 
thrush can employ a floater or ?wanderer? strategy (e.g. expanded home ranges) or 
large movements (Rappole et al. 1989, Winker et al. 1995). However, these studies 
were conducted at smaller spatial and temporal scales (transmitter life < 20 days), 
which makes characterizations of permanent versus temporary departures difficult. 
Although the probability of departure was high for both study populations, some 
individuals abstained from movements, suggesting this is not an obligate strategy. 
Furthermore, individuals that were tracked across multiple years on our Belize study 
site (n = 17), showed variability in seasonal site persistence across years (e.g. timing, 
site tenacity). These results suggest that non-breeding movements in wood thrush are 
a facultative strategy that allow for flexibility in space use among and within 
individuals. These results contribute to growing evidence that migratory birds engage 
in a wide array of both short- and long-distance movement behaviors during the non-
63 
 
 
  
breeding season that vary across species and populations of migratory songbirds 
(Stutchbury et al. 2016, Norevik et al. 2018). 
Facultative movements recorded from individuals tracked with GPS 
transmitters suggest that this behavior occurs primarily during the latter part of the 
non-breeding season (February - April). Wood thrush arrive on their non-breeding 
range at the end of the wet season (October - November), when insect abundance is 
high (Janzen and Schoener 1968) and as the dry season progresses insect abundance 
declines ( McKinnon et al. 2015). The timing of facultative movements suggest that 
the ultimate mechanism driving departure may be deteriorating environmental 
conditions, driven by declines in moisture (Terrill and Ohmart 1984). This behavior 
could be akin to the tracking of resources described in altitudinal (Boyle et al. 2010) 
and intra-tropical migration systems (Thorup et al. 2017, Kole?ek et al. 2018). For 
wood thrush, seasonal declines in moisture are more predictable and the scale of 
movements are smaller and similarly, birds may be initiating movements to search for 
improved environmental conditions. The broad range of departure dates suggests that 
individuals were not responding to broad-scale environmental cues (e.g. photoperiod, 
phenology; Heckscher et al. 2015, Thorup et al. 2017), but may instead be responding 
to finer-scale changes in habitat suitability. Therefore, facultative movements in wood 
thrush could be an adaptive response to fine-scale fluctuations in habitat suitability.  
Contrary to our predictions, individuals from wetter habitats had a higher 
probability of engaging in facultative movements. Furthermore, neither sex nor age 
class influenced the probability of departure. This suggests that facultative 
movements are not driven by competition avoidance by subordinate individuals (e.g. 
64 
 
 
  
competitive release; Smith and Nilsson 1987). Given the evidence that birds in wetter 
habitats and in better condition were more likely to engage in winter movements, we 
propose that facultative winter movements in wood thrush may be condition 
dependent (Ketterson and Nolan 1983). Therefore, as local habitat suitability 
declines, individuals in poor condition may be constrained from engaging in this 
putatively risky behavior (e.g. due to low fat or pectoral muscle scores) and may 
instead try to ?ride-out? deteriorating conditions. Given that birds must prepare for 
spring migration at the end of the dry season, individuals that attempt to ?ride-out? 
conditions may be forced to delay migration preparations should conditions not 
improve in time. Although we suggest that facultative non-breeding movements by 
wood thrush are ultimately a response to declining habitat suitability, we found no 
evidence that individual wood thrush were directly tracking seasonal resources 
(Thorup et al. 2017) or upgrading to higher quality habitats (Smith et al. 2011b). 
Based on GPS tracking, there was no difference in the overall average wetness or 
variability in wetness between locations pre- or post- movements. However, low 
sample size made it difficult to account for latitudinal and elevation differences 
among individuals. Furthermore, high cloud coverage in Central America makes 
time-series of fine-resolution satellite derived data difficult to obtain. Therefore, a 
large question left unanswered is the proximate mechanism driving timing of 
movement decisions and settlement in new locations. Additional factors that could 
drive habitat suitability and influence decision-making of individual birds, including 
interspecific competition (Stutchbury et al. 2016) or changes in predation risk, should 
also be considered. Comparing the fitness consequences (e.g. survival, condition, 
65 
 
 
  
reproductive success) of either strategy will be necessary to better resolve the value of 
facultative winter movements in wood thrush. 
Deciphering the role that seasonal patterns of moisture play for limiting 
populations of Neotropical migratory songbirds has important implications for the 
management of these species, as climate models forecast increasingly arid condition 
on their wintering range (Neelin et al. 2006). In this study, we found that individuals 
in drier conditions were in lower individual condition and adopted different space-use 
strategies, the consequences of which are still unclear. At local spatial scales, 
individuals in drier habitat were more likely to utilize larger home range. Previous 
research has suggested that wood thrush that occupy larger home ranges face higher 
mortality rates (?wanderers? sensu Rappole et al. 1989). At larger spatial scales, 
individuals in drier habitats were less likely to depart non-breeding home ranges as 
conditions deteriorated across the dry season. If facultative movements are an 
adaptive strategy to avoid deteriorating dry season habitat suitability, this strategy 
may buffer wood thrush against the predicted increased severity of dry seasons. 
However, if facultative movements are a condition dependent strategy, then 
increasingly dry conditions during the non-breeding period could prevent birds from 
escaping deteriorating conditions as the dry season progresses. Developing a better 
understanding of the mechanisms driving facultative movements of wood thrush, and 
perhaps other species of migratory birds, will be essential to understanding the 
flexibility individuals may need to exhibit to adjust to future climatic shifts.   
 
66 
 
 
  
Acknowledgements 
This work would not have been possible without the field assistance of M. Cruz, L. 
Duval, N. Freeman, J. Hightower, C. Hines, A. Sutton and L. Welsh in Belize. We 
also thank numerous field assistant at our field sites in the United States, notably T. 
Guida and P. Keller, as well as the volunteers from Forsyth Audubon, Bedford 
Audubon, Carpenter Natura Center, and Warner Nature Center for assistance at our 
field sites in the United States. We would like to thank J. Marlin, H. Barrett and J. 
Dourson for their logistical support at BFREE and K. Brand, Z. Landin, J. Robbins, 
P. Smithson, K. Thorington and J. Veith for their logistical support in the field in the 
United States. Funding for this study came from Strategic Environmental Research & 
Development Program (RC-2121), Minnesota National Park Service, Bedford 
Audubon Society, Forsyth Audubon Society, International Audubon Society and a 
United States Department of Agriculture McIntire-Stennis Capacity Grant and the 
BEES/BISI program. 
 
 
67 
 
 
  
Tables 
Table 1.1 Model averaged parameter estimates, standard error and unconditional 95% 
confidence intervals for top generalized mixed effect models for home range size 
(95% isopleth of utilization distribution) of non-breeding wood thrush in Belize. 
Separate models were constructed for isotope values derived from claws (n = 37) 
which provides information on diets over a few months and blood (n = 44) which 
provides information on diets over a few weeks. 
 
Parameter Estimate SE 95% CI 
Claw    
Age (yearling) -0.30 0.11 -0.51, -0.09 
Sex (Male) 0.32 0.13 0.07, 0.56 
    
Blood    
Age (yearling) -0.16 0.09 -0.34, 0.03 
Sex (Male) 0.10 0.10 -0.09, 0.30 
Condition 1.47 1.05 -0.60, 3.53 
?13C  -0.13 0.06 -0.26, -0.01 
?15N -0.05 0.06 -0.17, 0.06 
Notes: Values in bold have confidence intervals that do not include zero. 
 
68 
 
 
  
Table 1.2 Model averaged parameter estimates, standard error and unconditional 95% 
confidence intervals for top linear mixed effect models on the body condition index in 
the early dry season sampling period for non-breeding wood thrush in Belize. 
Separate models were constructed for isotope values derived from claws (n = 104) 
which provides information on diets over a few months and blood (n = 112) which 
provides information on diets over a few weeks. 
 
Parameter Estimate SE 95% CI 
Claw    
?13C -1.74 0.81 -3.32, -0.17 
Wetness -24.68 28.53 -80.59, 31.24 
Age (yearling) 2.31 1.72 -1.07, 5.68 
Sex (Male) 3.76 1.69 0.45, 7.07 
Wetness * ?13C 38.34 17.61 3.83, 72.86 
Age (yearling) * Sex (Male) -6.33 2.37 -10.98, -1.69 
?15N 1.61 0.77 0.11, 3.11 
Wetness * ?15N -19.98 17.05 -53.4, 13.44 
    
Blood    
?13C -2.18 0.66 -3.48, -0.88 
?15N 2.37 0.91 0.59, 4.15 
Wetness -7.84 27.46 -61.66, 45.98 
Age (yearling) 2.25 1.68 -1.03, 5.54 
Sex (Male) 4.26 1.71 0.9, 7.62 
Age (yearling) * Sex (Male) -5.42 2.22 -9.77, -1.07 
Wetness * ?15N -39.93 23.65 -86.29, 6.43 
Wetness * ?13C 30.08 17.00 -3.25, 63.41 
Notes: Values in bold have confidence intervals that do not include zero. 
69 
 
 
  
Table 1.3 Model averaged parameter estimates, standard error and unconditional 95% 
confidence intervals for top linear mixed effect models describing change in mass 
between the early and late dry season sampling period for non-breeding wood thrush 
in Belize. Separate models were constructed for isotope values derived from claws (n 
= 26) which provides information on diets over a few months and blood (n = 31) 
which provides information on diets over a few weeks. 
Parameter Estimate SE 95% CI 
Claw    
?13C 0.68 0.60 -0.5, 1.87 
Wetness 14.21 40.86 -65.87, 94.28 
Age (yearling) 0.82 1.43 -1.98, 3.63 
Sex (Male) 1.22 1.32 -1.37, 3.82 
Age (yearling) * Sex (Male) 0.13 1.94 -3.68, 3.94 
?15N -0.22 0.53 -1.26, 0.81 
    
Blood    
?13C -1.11 0.53 -2.15, -0.07 
Wetness 2.36 19.16 -35.2, 39.91 
Age (yearling) -0.52 1.42 -3.3, 2.26 
Sex (Male) 0.07 1.21 -2.31, 2.45 
Age (yearling) * Sex (Male) 1.80 1.91 -1.94, 5.55 
?15N -0.79 0.45 -1.68, 0.1 
Notes: Values in bold have confidence intervals that do not include zero. 
 
70 
 
 
  
Figures 
Figure Legend 
Figure 2.1. Factors driving home range size (A ? C), food availability (D ? E) and 
food consumption (F) for wood thrush during the non-breeding season. Mean (? SE) 
home range size (ha) for each age (A) and sex (B) class of wood thrush. (C) Positive 
relationship between home range size and habitat moisture based on stable-carbon 
values from claws. Stable-carbon values have been scaled; more negative values 
denote drier habitats. Fruit (D) and insect (E) counts conducted on individual wood 
thrush home ranges were positively related to satellite-derived home range wetness. 
Fruit and insect counts were log-transformed and scaled, more positive wetness 
values denote wetter habitats. (F) Negative relationship between stable-nitrogen 
levels derived from claws and satellite-derived plot wetness. Stable nitrogen values 
have been scaled; more positive values denote larger proportions of insects in diet. 
More negative wetness values denote drier conditions. Asterix denotes significant 
difference in bar plots. For scatterplots, line denotes the model predicted relationship 
and shaded area the 95% CI.  
 
Figure 2.2. Factors driving body condition of wood thrush during the non-breeding 
season in Belize. (A) Body condition index (marginal mean ? 95% CI) during the 
early dry season by age and sex class. Body condition index (B) and change in mass 
(D) were negatively related to stable-carbon values in the blood. Stable-carbon value 
have been scaled, more negative values denote drier habitats. (C) Positive relationship 
71 
 
 
  
between stable-nitrogen values in blood and body condition index. Stable nitrogen 
values scaled, more positive values denote larger proportions of insects in diet. For 
scatterplots, line denotes the model predicted relationship and shaded area the 95% 
CI. 
 
Figure 2.3. Proportion of wood thrush remaining stationary across the non-breeding 
period from our study population in Belize and wood thrush tracked using GPS 
transmitters based on the estimated survival function of a Cox Proportional Hazard 
model. (A) In Belize, estimated proportion of wood thrush remaining stationary in 
wet (solid, red line) and dry (dashed, black line) habitats based on stable-carbon 
values in claws. Enriched ?13C (expressed as first quartile) represents wetter habitats, 
depleted ?13C (expressed as third quartile) represent dry habitats. (B) Negative effect 
of habitat moisture as measured by ?13C values in claws on the hazard ratio for 
departure. The probability of performing a facultative movement increased under 
wetter conditions, represented by more negative (depleted) ?13C values. (C) For GPS 
tracked individuals, proportion of wood thrush remaining stationary in wet (solid, red 
line) and dry (dashed, black line) habitats based on satellite-derived wetness values. 
Dry values are expressed as the first quartile of wetness values, wet values are 
expressed as the third quartile of wetness values. (D) Positive effect of habitat 
moisture as measured by the satellite-derived wetness value on the hazard ratio of 
departure. The probability of engaging in facultative movements increases with 
higher wetness values (wetter). (A, C) Line represents estimate of proportion of 
72 
 
 
  
stationary individuals from model, shaded area 95% CI. (B, D) Line represent 
adjusted hazard ratio and shaded area represents 95% CI. 
 
Figure 2.4. (A) Non-breeding locations (bi-monthly) of adult male wood thrush 
tracked using high-resolution GPS transmitters from 5 breeding populations (n = 17). 
Circles represents individuals that remained stationary across the non-breeding 
season. Triangles represent individuals that made non-breeding movements. Different 
combinations of triangle type (delta vs. inverse) and shades represent unique 
individuals. (B) Non-breeding movements of 4 individuals in Honduras and 
Nicaragua, lines connect individual movements. (C) Timing of intra-winter 
movements for two individuals in Nicaragua. 
73 
 
 
  
Figure 2.1 
 
 
 
74 
 
 
  
Figure 2.2 
 
 
75 
 
 
  
Figure 2.3 
 
 
76 
 
 
  
Figure 2.4 
 
 
77 
 
 
  
 : Food limitation and the endogenous control of 
spring migration: a captive experiment with a long-distance 
migratory bird  
 
Abstract 
The onset of spring migratory behavior in birds is thought to be controlled by a rigid 
endogenous schedule due to the selection pressure to time migration in anticipation of 
environmental conditions on the breeding grounds. The degree of flexibility for 
spring departure schedules in response to supplementary environmental cues remains 
unclear.  Recent field studies have suggested an important role of external factors 
such as weather variation and food availability in modifying the timing of spring 
departures.  Here, we investigated the role of food availability in modifying 
endogenous spring migratory schedules in a captive population of wild-caught Wood 
Thrush (Hylocichla mustelina) by quantifying nocturnal migratory restlessness 
(Zugunruhe) under two food treatments: ad libitum or ?control? food availability 
expected to maintain body weight vs. reduced food availability expected to result in a 
5% decline in body weight. Contrary to predictions, food treatment did not affect the 
onset of Zugunruhe, however, it did advance migratory fattening compared to the 
control group. Birds with reduced food availability maintained lower fuel loads (% 
target mass) and decreased Zugunruhe. Moreover, fuel load predicted the cumulative 
magnitude of Zugunruhe. These results suggest that food availability may modulate 
the development of Zugunruhe, not its onset, in captive birds adding support to field 
studies that have demonstrated delayed spring migration departure in physically 
compromised birds. Fattening when food availability is low may provide a strategy to 
78 
 
 
  
escape poor or unpredictable winter environmental conditions regardless of migratory 
readiness.  Our results highlight the challenging trade-off faced by migratory birds 
between surviving migration and optimally timing arrival on the breeding ground and 
further solidifies how periods of the annual cycle are linked and the significant role 
these interacting effects can play in driving the fitness of migratory birds. 
 
Introduction 
Migratory species are expected to time their migrations optimally to match the timing 
of life history events with suitable environmental conditions (Alerstam and Lindstr?m 
1990). For migratory birds, this is expected to be particularly important during spring 
migration because even relatively small differences in the timing of arrival and 
initiation of nesting on breeding grounds can have profound fitness consequences 
(Saino et al. 2004, Reudink et al. 2009).  To correctly time migration relative to 
anticipated environmental conditions on northern breeding areas, migratory birds are 
thought to rely on an endogenous circannual schedule synchronized to photoperiod 
that acts as a predictable environmental cue (zeitgeber; Gwinner 1996, ?kesson et al. 
2017). To avoid the risk of mistiming migration, a rigid endogenous schedule would 
be favoured since additional environmental stimuli on the nonbreeding grounds will 
have limited or no predictive relevance to current conditions on breeding sites.  At the 
same time, to maximize survival during migration, selection may favor individuals 
that can flexibly adjust rigid timing mechanisms to respond to current environmental 
conditions (Helm et al. 2017). For example, spatial (e.g. Latta and Faaborg 2002, 
Studds and Marra 2008) and temporal (e.g. Studds and Marra 2007, McKinnon et al. 
79 
 
 
  
2015) variation in non-breeding habitat suitability have been found to influence 
individual non-breeding performance. To adjust to these fluctuations in non-breeding 
conditions, other factors, such as temperature, precipitation and food availability, may 
impact migration progression by either accelerating or deaccelerating migratory 
function (Ramenofsky et al. 2012). Mechanistically, precipitation and moisture 
availability have been shown to control food (arthropod) availability, which limits the 
physical condition of overwintering American Redstarts (Setophaga ruticilla; Studds 
and Marra 2007). Whether precipitation and food availability act directly as a cue, or, 
indirectly as a constraint on migration preparation (e.g. migratory fattening) to fine-
tuning departure decisions is unknown. Regardless, a quantitative understanding of 
the role of these factors in influencing the different phases of migration is currently 
lacking for most species.  
The migratory period can be divided into three phases: development, mature 
expression and arrival sensu Cornelius et al. (2013). The developmental phase 
involves the physiological transition to migratory disposition; mature expression is 
characterized by consecutive cycles of fuelling and flight; and arrival involves the 
termination of migratory behaviour and transition to subsequent life history stages. 
Most of what is known about the proximate mechanisms underlying the timing of 
migration comes from experiments on captive birds (Gauthreaux, 1996). Under 
captive situations most obligate migratory species develop fat reserves and exhibit 
nocturnal restlessness or Zugunruhe, typified by periods of wing-whirring, increased 
perch-hopping, body-turning and other activities. The intensity of these behaviours 
provide a simple metric to quantify the disposition of captive birds to migrate in the 
80 
 
 
  
wild (e.g. birds exhibiting lower intensity of Zugunruhe have lower migratory 
disposition). Currently the best understanding of the role of external factors in 
modifying the onset of migration comes from studies of birds during the mature phase 
of migration. Studies of free-living individuals have found that departure decisions 
from stopover sites are influenced by atmospheric, climatic and physiological 
conditions (Cochran and Wikelski 2005, Fusani et al. 2011) while termination of 
migration has been linked to climate and availability of food resources (Terrill and 
Ohmart 1984, Ramenofsky and Wingfield 2017). The role of food availability and 
physiological condition has been extensively examined due to the importance of 
acquiring sufficient fat reserves to fuel migration (Bairlein 2003).  
Studies examining migrants at stopover locations or by stimulating stopover 
conditions in captivity (fasting-refueling experiments) have generally found that (1) 
fasting increased migratory activity and (2) refueling decreased migratory activity 
(Gwinner et al. 1988, Eikenaar et al. 2014), but that these effects were mediated by 
physiological condition. Individuals in poor condition (or who had lost large fuel 
loads) were less likely to leave stopover sites in field studies or displayed lower 
Zugunruhe in captive experiments (Bridge et al. 2010, Fusani et al. 2011). Together 
these studies suggest that the migration program does show flexibility and 
supplemental environmental factors can provide spatiotemporally sensitive 
information to allow fine-tuning of departure decisions based on current behavioural 
and physiological condition. The degree of flexibility likely varies across species 
depending on a number of factors including life-history stage (spring vs fall), 
81 
 
 
  
migratory strategy (facultative vs obligate), and ecological conditions (habitat type, 
diet).   
How supplementary environmental factors influence the development phase 
of spring migration on the wintering grounds is less well resolved. An emphasis on a 
rigid endogenous departure schedule was bolstered by early work suggesting a minor 
role for food availability as a local supplementary cue promoting the development of 
migratory activity (Berthold 1975, Lofts et al. 2008). However, this work was 
restricted to fall migratory activity in captive individuals. Far less research has been 
conducted examining the effects of supplemental environmental factors on spring 
departure decisions. Recent work investigating the ecology of free-living populations 
of migratory songbirds on their non-breeding grounds has demonstrated how winter 
habitat quality, often linked to precipitation, can influence spring departure, shedding 
new light on how seasonal interactions drive migratory function (Marra et al. 1998, 
Studds and Marra 2011b, McKinnon et al. 2015b). Collectively, these studies in wild 
populations suggest that factors, such as food availability, can delay the circannual 
time program (Helm et al. 2017).  On average, these studies observe differences of 3-
5 days between years, suggesting flexibility in departure possibly linked to 
environmental and physical conditions in both observational (Studds and Marra 
2011b) and tracking studies (Stanley et al. 2012, Conklin et al. 2013). Therefore, the 
role supplementary factors play in modulating the rate and magnitude of the 
development of spring migratory behaviour is unknown. The degree of plasticity is 
likely a fine balance between the benefits of ensuring optimal timing of arrival at 
82 
 
 
  
breeding grounds and the ability to flexibly adjust to current conditions to maximize 
survival while on migration. 
To experimentally examine the interaction between the endogenous migratory 
control program and supplementary factors, such as food availability, we performed a 
controlled feeding experiment on captive Wood Thrush (Hylocichla mustelina) and 
monitored their spring migratory activity.  Specifically, we examined if and how food 
availability regulates the onset and intensity of Zugunruhe and migratory fattening. 
We predicted that birds on a food-restricted diet would have decreased fuel load, 
leading to later onset of spring migratory activity and decreased Zugunruhe compared 
to controls with food available ad libitum.  
 
Methods 
Study System 
Wood Thrushes are obligate long-distance Neotropical migratory songbirds that breed 
in eastern North America and spend the non-breeding season in Central America and 
Mexico. Tracking studies have revealed some degree of repeatability in spring 
departure dates for individual Wood Thrush tracked over multiple years, with an 
average difference of 3 days between years (Stanley et al. 2012). McKinnon et al. 
(2015b) identified an association between departure dates and dry environmental 
condition based on coarse resolution remote sensing data (NDVI), but no association 
with late season body condition. Therefore, the Wood Thrush migratory schedule 
may show some flexibility in departures dates, but it is unclear whether food 
83 
 
 
  
availability or some other physical change on the winter grounds drove these 
differences. 
Bird Capture and Maintenance 
To investigate the role food availability plays in modifying spring departure dates, 
adult male Wood Thrush (n = 26) were captured using mist nets in the summer of 
2015 on Maryland Department of Natural Resources land units in Carrol, Frederick, 
Montgomery and Prince Georges counties. Birds were banded, body measurements 
taken, and transported to the Smithsonian National Zoological Park research facilities 
where they were housed in individual cages (66 x 36 x 50 cm) kept on a 16L:8D 
photoperiod and a constant temperature (~26oC). They were switched to a 12L:12D 
light cycle to stimulate the overwinter period on 5 March 2016. Birds were 
maintained on an agar-based mash diet designed to provide 78 kcal and were fed 
twice per day. The diet consisted of Zulife Softbill Diet (1.8%), Proplan Large Breed 
Dog (1.8%), fruit and vegetable mix (apple, banana, cantaloupe, papaya, zucchini, 
blueberries; 15.2%), hard-boiled egg (6.6%), insects (crickets, earthworms, 
mealworms and waxworms; 25.5%), agar (1.2%) and water (38.3%), and topped with 
2 g of live mealworms.  
Body Condition 
To monitor the experimental treatments, body condition was recorded twice per week 
in the morning prior to individuals receiving food. Measurements included body mass 
(nearest 0.1g), and subcutaneous fat deposits. To measure migratory fuel reserves, 
fuel load was determined using the formula: (body mass ? target winter body 
84 
 
 
  
mass)/target winter body mass. Target winter body mass was estimated for each bird 
based on their structural size using a regression equation of mass versus metatarsus 
(to the nearest 0.1 mm) and wing chord (to the nearest 1 mm) generated from a 
dataset (Chapter 2) for Wood Thrush wintering in Belize with low subcutaneous fat 
scores (trace or 1 fat; 3 years of data; mass = 18.5 + 0.31(metatarsus) + 1.03(wing 
chord). Fat deposits were scored by examining the torso visually using a standardized 
5-point scoring scale (Holmes et al. 1989).  
Nocturnal Activity 
To monitor Zugunruhe, nocturnal activity was recorded every night using infra-red 
video cameras (Swann 1080p Digital Video Recorder and Cameras, Model # 
SWDVK-845008) placed above cages. Each camera recorded the movement of 4 
cages simultaneously (Figure 3.1b). Videos were analyzed using BirdOriTrack 
software (Muheim et al. 2014). Nightly nocturnal activity was calculated as the sum 
of the total length moved every 30 seconds relative to the length of the cage. 
Nocturnal activity was calculated from 30 minutes after lights out until 30 minutes 
prior to lights on.  
Experimental design  
We randomly assigned 13 birds to the food reduction treatment and 13 birds to the ad 
libitum treatment (Figure 3.1a). The ad libitum treatment was designed to maintain 
target body weight, whereas the food reduction treatment was designed to decrease 
target body weight by 5%. The food reduction treatment was achieved by reducing 
the standard diet provided to each bird. The reduction ranged from - 55% to +15% 
85 
 
 
  
depending on the individual, but the composition of the diet remained consistent. 
After three weeks individuals reached the desired body mass for their experimental 
feeding treatment and were kept on the experimental treatment until the end of the 
experiment. To promote the transition into migration, the photoperiod was gradually 
increased over 24 days from 12L:12D to 13L:11D to mimic late-winter pre-migratory 
daylength in the natural environment (Figure 3.2a). To monitor the development of 
migratory behaviour we measured: (1) Zugunruhe and (2) migratory fattening.  
Timing of Migration  
To identify periods of Zugunruhe changepoint analysis was applied to the time series 
data of average daily nocturnal activity using the binary segmentation algorithm in 
the R package changepoint (Doren et al. 2017). Changepoint analysis identifies the 
point in time when the statistical properties (e.g. mean, variance) of a sequence of 
data changes (Killick and Eckley 2015). We used the ?cpt.mean? function which 
detects changes in the mean of a time series and a cumulative sum test statistic, which 
does not assume a distribution for the data. To define the Zugunruhe period we 
restricted the number of changepoints to two with a minimum segment length of 20 
since we expected Zugunruhe periods to be weeks in duration (Doren et al. 2017).  
Onset and termination of migration were defined as the start and end dates identified 
by changepoint analysis as periods of elevated nocturnal activity, duration was the 
number of days between the start and end dates. Birds were excluded from analysis if 
no changepoint was identified (e.g. consistent low or high nocturnal activity; n = 7), 
or if changepoints were detected prior to the commencement of the experimental food 
86 
 
 
  
treatment (n = 3). In total 10 birds were omitted from analyses (5 from each treatment 
group), resulting in a final sample size of 16 birds (n = 8 per group).  
To identify periods of migratory fattening (i.e. the accumulation of fat as an 
energy reserve for migratory flights) we identified periods of increased fuel load 
using the same methods as Maggini and Bairlein (2010). In brief, we calculated mean 
fuel loads in five-day bins and identified periods of 3 consecutive increases or 
decreases. The third day of the first bin of increase was identified as the onset of 
migration. The end of migration was calculated by the same method but in reverse 
direction (3 consecutive decreases, third day of last bin).  
Magnitude of Migration 
The magnitude of Zugunruhe was calculated using a cumulative measure of 
Zugunruhe calculated by measuring the area under the curve formed by taking the 
average nocturnal activity at each day across the (i) migratory period defined by 
changepoint analysis and (ii) migratory fattening period. Area under the curve was 
calculated using linear interpolation with the auc function in the R package MESS 
(Ekstr?m 2019). 
Statistics  
To determine the influence of feeding treatment on body condition we examined fuel 
load and fat scores at two time periods: prior to migration and across each bird?s 
migratory period based on changepoint analysis of Zugunruhe data. For each time 
period we compared treatment groups using generalized linear models.  To examine 
the effect of feeding treatment on timing and magnitude of migratory behaviour we 
87 
 
 
  
constructed separate generalized linear models for each timing and magnitude metric 
(e.g. onset of migratory fattening, average intensity of Zugunruhe) to compare 
treatment groups. To examine how body condition influenced the magnitude of 
Zugunruhe we compared cumulative Zugunruhe to average fuel load using 
generalized linear models. 
All analyses were conducted in R 3.4.3 (R Core Team 2017). Generalized 
linear models were run in the R package stats (R Core Team 2017). To assess model 
fit, each model was compared to a null model using an F-test. All models were 
assessed for violation of model assumptions by visual inspection of plots of residual 
versus predicted values and quantile-quantile plots. When necessary to meet model 
assumptions of normality response variables were transformed and significant outliers 
were identified using Cook?s distance plots (Kutner et al. 2004). One bird was 
excluded from models of Zugunruhe magnitude as an outlier (Cook?s distance > 1). 
 
Results 
Body Condition 
Prior to the start of food treatments there were no differences in fuel load or fat scores 
between treatment groups (Fuel load: F1,16 = -0.0554, p = 0.114; Fat: F1,16 = -0.0542, p 
= 0.896). Food reduction decreased condition compared to the control group (Table 
3.1). At the beginning of the experimental trial, prior to the development of 
Zugunruhe, fuel loads were lower in the reduction food treatment, but not 
significantly. (F1,13 = -0.04106, p = 0.0566). There was no difference in fat scores 
88 
 
 
  
(F1,13 = -0.171, p = 0.646). Once migratory condition began all birds put on fat 
deposits and gained weight, irrespective of treatment. During Zugunruhe expression, 
food reduction birds exhibited significantly lowered fuel loads (F1,15 = -0.147, p = 
0.0030) and fat scores (F1,15 = -1.030, p = 0.0353) compared to the control birds.  
Timing of Migration 
Birds began developing migratory disposition (either migration fattening or 
Zugunruhe) between 1 to 6 weeks after the start of the experimental feeding 
treatments (Julian dates range: 67 to 107; Figure 3.2). Feeding treatment had no effect 
on the start or end day of Zugunruhe (start: F1,14 = 0.0443, p = 0.446; end: F1,14 = -
0.0446, p = 0.287). Food restriction advanced the start of migratory fattening by 2-3 
days (F1,16 = -3.097, p = 0.0276; Figure 3.2) but had no effect on the termination day 
of migratory fattening (F1,14 = -0.0446, p = 0.287). Overall there was a broader range 
in onset of Zugunruhe (72 days, range = 67 to 145) compared with migratory 
fattening (9 days, range = 98 to 107). Most individuals began exhibiting Zugunruhe 
prior to migratory fattening; however, four individuals (Food reduction: n = 3; 
Control: n = 1) did not exhibit Zugunruhe until after migratory fattening. Six 
individuals began developing Zugunruhe prior to the increase in photoperiod, three 
from each treatment group.  
Magnitude of Migration 
Across the migratory period, food restricted birds exhibited lower cumulative 
Zugunruhe, although it was not significant (F1,14 = -9.812, p = 0. 0796; Figure 3.3). 
Across the migratory fattening period, food restriction significantly decreased 
89 
 
 
  
cumulative Zugunruhe (F1,14 = 0.0381, p = 0.0502). Fuel load had a significant 
positive effect on cumulative Zugunruhe across the migratory period (F1,14 = 80.182, 
p = 0.0415) and the effect approached significance across the migratory fattening 
period (F1,14 = -0.260, p = 0.0558; Figure 3.4).  
 
Discussion 
To ensure optimal timing of reproduction, long-distance migratory songbirds have 
been thought to rely on a rigid genetically controlled spring migration schedule to 
ensure timely departure from wintering grounds (Berthold et al. 2003). The degree of 
flexibility to fine tune timing of winter departure, based on supplementary 
environmental factors such as food availability, remains poorly resolved but could 
represent an important strategy for ensuring successful completion of migration. Here 
we found that reduced food availability lowers migratory body condition and 
decreases a cumulative measure of Zugunruhe. However, we did not identify 
differences in the timing of expression of Zugunruhe between diets, but instead found 
that onset of migratory fattening was advanced in individuals on reduced diets. Our 
study provides the first direct evidence that food availability modulates the 
development and intensity of spring migratory behaviour in an obligate long-distance 
migratory songbird.   
 Food availability is expected to be an important supplementary factor because 
the development of mature migratory capabilities relies on the accumulation of fuel 
stores (Ramenofsky et al. 2012, Cornelius et al. 2013). Despite the role of fat stores in 
fueling migration, early studies on captive migrants did not consistently find an effect 
90 
 
 
  
of food restriction on migratory behaviour in captive birds. More recent research 
examining later phases of migration (mature expression) have found that fuel load 
and/or amount of fuel lost is a good predictor of intensity of Zugunruhe (Fusani et al. 
2011, Eikenaar et al. 2014). In this study, by challenging individuals as they 
developed migratory disposition, we similarly observed inhibition of Zugunruhe as 
birds attempted to accumulate fat reserves on a reduced diet and positive effects of 
fuel load on Zugunruhe. If intensity of Zugunruhe represent the disposition of a bird 
to depart on migration, we interpret the inhibition of cumulative Zugunruhe to 
suggest that low food availability on the non-breeding grounds will slow progression 
of migration development and possibly delay departure on migration. These results 
suggest that inter-annual variation in spring departure schedules, as observed in 
American Redstarts (Studds and Marra 2011b) and Wood Thrush (Stanley et al. 
2012) by monitoring or tracking the same individuals across years, could be 
explained by differences in food availability on wintering territories. Flexibility to 
adjust the timing program based on non-breeding ground condition could therefore 
provide an important mechanism to handle variability in habitat suitability and 
maximize survival on migration. However, if birds are not able to correct for delays 
(e.g. faster migration; McKinnon et al. 2015b), events on the wintering grounds 
would likely carry-over to influence timing of subsequent life-history stages and 
reproductive success (Marra et al. 1998, Reudink et al. 2009).  
Contrary to expectations, we found food reduction advanced onset of 
migratory fattening an average of 3 days. Attempts to prevent migratory fattening 
were unsuccessful and accumulation of migratory fat reserves occurred in the food 
91 
 
 
  
reduction group, likely through increased food utilization efficiency and not 
hyperphagia as birds did not have access to extra food (Bairlein 2003). During fall 
migration, intermittent fasting similarly advanced fall migratory fattening in garden 
warblers (Sylvia borin; Totzke et al. 2000). The advancement of migratory fattening 
could be a strategy to buffer birds against unpredictable environmental conditions 
ahead of migration.  For example, increased fat reserves has been predicted and 
observed in prior experiments as a form of insurance during periods of temporary 
nutritional stress  (McNamara and Houston 1990, Gosler 1996). Alternatively, low 
food availability, resulting in a negative energy balance, may also provide a 
supplemental cue to the migratory fattening program and serves to accelerate fuel 
accumulation. If birds require a threshold fuel load for departure, advancing 
migratory fattening could afford additional time to accumulate sufficient fat reserves. 
An interesting possibility might be advanced based on findings from tracking studies 
that revealed that Wood Thrush engage in facultative movements during the winter 
period (Chapter 2). Accelerating migratory development could therefore provide 
individuals the opportunity to locate staging areas more suitable for migration 
preparation prior to spring departures. Since low food availability inhibits the 
intensity of Zugunruhe, the ability to advance migratory fattening may provide a 
mechanism to avoid delays in the spring migratory program by advancing preparation 
for departures.  
The rigidity of spring migration schedules is often espoused due to (1) the 
fitness consequences associated with mistiming arrival on the breeding grounds and 
(2) the lack of relevant environmental factors on non-breeding areas to predict 
92 
 
 
  
breeding conditions. However, a growing body of work suggests that obligate 
migrants show facultative components with their migrations, particular during later 
phases of migration (Ramenofsky et al. 2012).  For example, in extreme sub-alpine 
conditions, Gambel?s white-crowned sparrows (Zonotrichia leucophrys gambelii) 
arrive on breeding sites when snow still blankets the ground (Ramenofsky and 
Wingfield 2017). To balance the reproductive advantage of early arrival with the 
survival challenge of cold temperate breeding grounds, should conditions deteriorate, 
individuals are thought to engage in facultative irruptive flights by extending the 
migratory state until conditions improve. Additionally, during fall migration, both 
short- and long-distance migrants have been shown to redevelop migratory 
disposition at the end of migration, sometimes extending well into the winter period, 
if environmental or social conditions deteriorate (Berthold 1978, Terrill and Ohmart 
1984). The results presented here suggest that onset of spring migration may also 
display a facultative aspect most often associated with later phases of migration. This 
may be particularly relevant for species with unpredictable late-winter environmental 
conditions. For example, many Neotropical migratory birds prepare for spring 
migration at the end of the tropical dry season which vary in severity depending on 
yearly rainfall (Studds and Marra 2011b). Being able to advance migration 
preparation in dry years could prevent delays from influencing subsequent activities 
through seasonal interactions.   
The hypothesis that lower food availability would delay onset of Zugunruhe 
was not supported by the present study, which suggests the temporal control of 
Zugunruhe onset and migratory fattening are independent. In general, there was a 
93 
 
 
  
separation between migratory fattening and onset of Zugunruhe and more variation in 
onset of Zugunruhe (range = 107 days) compared with migratory fattening (range = 9 
days). This separation has been found in other studies and it is thought that these 
processes evolved independently (Berthold 1975, Lofts et al. 2008) and may be in 
part attributable to each bird?s genetically determined endogenous circannual 
schedule (Helm et al. 2017).   
The unnatural conditions the birds were housed in may have additionally 
contributed to this variation (unnatural photoperiod, proximity of birds to one 
another, missing environmental factors; Calisi and Bentley 2009). It is unclear 
whether migration departure is best represented by onset of Zugunruhe or a threshold 
intensity of Zugunruhe. In general, the intensity of Zugunruhe corresponds well to 
departure decisions in wild counterparts (Alerstam and Lindstr?m 1990, Eikenaar et 
al. 2014). In a recent tracking study of European Blackbirds (Turdus merula), a short-
distant migrant, individuals did not shift diel activity patterns until night of departure 
(Z??iga et al. 2016). Therefore, whether onset of Zugunruhe versus achieving a 
threshold intensity of Zugunruhe represents a departure decision in captive settings 
remains unclear. Tentatively, these results could be interpreted to suggest that 
development of Zugunruhe may be less (or not) responsive to food availability, 
compared with onset of migratory fattening, but disposition to depart as measured 
through intensity of activity is dampened by low food availability.  
This study provides novel quantitative evidence that support the hypothesis 
that low food availability during spring migration development could reduce 
cumulative Zugunruhe and can advance migratory fattening in captive Wood Thrush. 
94 
 
 
  
Therefore, the availability of food on wintering territories prior to spring migration 
could play a role in modulating the spring migratory program. Low food availability 
could therefore delay spring migration schedules, if birds are not able to develop 
sufficient fuel reserves, as an adaptive strategy to maximize survival during the 
migration period. These results support prior studies of wild populations of wintering 
songbirds which have found an association between winter habitat quality, spring 
departure dates and reproductive success (Rockwell et al. 2012, Saino et al. 2017). 
For example, field studies have shown that American Redstarts wintering in low 
quality habitat and in poor physical condition depart later on spring migration, 
resulting in more variable arrival timing on breeding grounds and reduced breeding 
success (Marra et al. 1998, Reudink et al. 2009). When individuals are experimentally 
upgraded to higher quality non-breeding habitats and improve their physical 
condition, they advance departure dates on spring migration compared to controls in 
lower quality habitat (Studds and Marra 2005). Results from studies on American 
Redstarts suggest that birds modify departure timing depending on non-breeding 
environmental conditions and the physical condition of the birds themselves. 
Together, this work also highlights the tricky timing challenge faced by migratory 
songbirds between migration survival and optimal breeding ground arrival. The 
advancement of spring migration fattening observed here could represent an 
important strategy to allow Wood Thrush to flexibly adjust migration timing when 
faced with unfavourable environmental conditions ahead of spring migration.  
The interaction of individuals with their environment can be difficult to 
monitor in highly mobile species; therefore, controlled experiments like ours isolate 
95 
 
 
  
the potential roles and interactions among behaviour, food availability and timing of 
spring migration. Flexibility in response to supplemental factors could have important 
implications for how well migratory birds adjust to climatic variation in the future 
(?kesson et al. 2017). As climate change increases and tropical deforestation 
proceeds, the wintering grounds of Wood Thrush and many Neotropical-Nearctic 
migrants are expected to experience increasingly severe dry seasons, which are 
predicted to decrease food availability for wintering birds (Neelin et al. 2006). The 
ability to modulate spring migration programs and advance migratory fattening could 
provide Wood Thrush flexibility to respond to changing environmental conditions 
before migration departure is delayed. However, advancements in migratory fattening 
may ultimately be limited if advancements in the other programs (i.e. Zugunruhe) 
associated with the vernal migratory stage are not equally plastic. 
 
Acknowledgements 
We thank Adrienne Dale, Kevin Bennett, Justin Saunders, Rebecca Zurlo for 
assistance with the collection and husbandry of the Wood Thrush collection. We 
would also like to thank all the FONZ and University of Maryland volunteers, as well 
as the National Zoological Park staff who aided with animal husbandry, most notably 
Elizabeth Fisher, Erin Kendrick, Mike Maslanka, Eric Slovak, Debra Talbot, Jordana 
Todd, and Jennifer Wall.  
96 
 
 
  
Tables 
Table 1.1 Mean fuel load and fat scores between individuals in Ad Libitum and 
Reduction groups prior to migration and during migration. Migratory period was 
defined for each individual based on the onset of migratory restlessness. Values in 
bold denote a significant difference (p < 0.05) and a tilde (~) a difference that 
approaches significance (p < 0.06) between the Ad Libitum and Reduction feeding 
treatments. 
 
 Ad Libitum Reduction 
 n Mean ? SE n Mean ? SE 
Fuel Load     
Pre-migration 7 0.0457 ?0.0138 7 ~ 0.00466 ? 0.0137 
Migration 8 0.155 ? 0.024 8 0.00738 ? 0.0332 
     
Fat Scores     
Pre-migration 7 1.47 ? 0.288 7 1.30 ? 0.220 
Migration 8 2.82 ?0.766 8 1.79 ? 0.350 
 
97 
 
 
  
Figures 
Figure Legend 
Figure 3.1. (A) Experimental protocol: birds were randomly assigned to one of two 
feeding treatments Ad Libitum or Reduction (n = 13 each) and maintained at a 
constant 12L:12D to mimic tropical winter conditions. After 3 weeks, birds were 
exposed to a changing photoperiod with daily shifts to longer light periods to 
photostimulate birds and mimic spring migratory conditions. (B) Image of cage and 
video camera set up. Each infra-red camera recorded activity from four cages 
simultaneously.  
 
Figure 3.2. Timing of spring migration by feeding treatment. (A) Shows the change in 
daylength across the experimental period. (B) and (C) boxplots show the start and end 
day of migration based on (B) changepoint analysis and (C) migratory fattening. 
Asterix represents significant (p < 0.05) differences between feeding treatment. 
 
Figure 3.3. Cumulative intensity of Zugunruhe by feeding treatment compared across 
the migratory period defined by changepoint analysis and the migratory fattening 
period. Asterix represents significant (p < 0.05) differences between feeding 
treatment and a tilde (~) a difference that approaches significance (p < 0.08). 
 
Figure 3.4. Effect of mean fuel load across the migratory period on cumulative 
Zugunruhe during the migratory period defined by (A) changepoint analysis and (B) 
98 
 
 
  
migratory fattening. Treatment groups shown as reference (filled: Reduction, open: 
Ad Libitum). 
99 
 
 
  
Figure 3.1 
 
 
 
 
100 
 
 
  
Figure 3.2 
 
 
 
101 
 
 
  
Figure 3.3 
 
 
102 
 
 
  
Figure 3.4 
 
 
103 
 
 
  
Appendices 
 
Supplement to Chapter 1: 
 Supplementary Methods 
 Supplementary Tables S1-S8 
 
Supplementary Methods 
 
Geolocator Deployment and Programming 
 
Individual wood thrush were captured using mist-nets and a song playback.  Upon 
capture identification bands were placed on their legs (one aluminum United States 
Geological Service band and up to three plastic color bands in unique combinations) 
and standard body measurements taken (mass, fat score, tarsus, wing, bill size). GPS 
tags were attached using a backpack leg-loop harness (<5% of birds body mass; 
Rappole and Tipton 1991). We preferentially tagged adult males because they have 
higher site fidelity and to reduce confounding variables due to age and sex specific 
migratory schedules (Stanley et al. 2012). Recapture rates were variable among sites, 
the average recapture rate for tagged individuals was 0.16 ? 0.10% (?SD), which 
approached significantly lower levels than the recapture rate of untagged males (0.24 
? 0.14%, tpaired = -1.73, df = 8, p-value = 0.06). No significant difference in the mass 
of birds between deployment and retrieval of tags was detected (t = 0.73, df = 16, p = 
0.81).  
To maximize the number of location fixes per season, GPS tags were 
programmed to obtain fixes every 2 weeks during stationary periods, every 4 days 
during fall migration and every 3 days during spring migration. Fixes were obtained 
at midnight EST during stationary periods (roosting birds have higher probability of 
104 
 
 
  
communication with satellites) and at noon EST during migratory periods to capture 
stopover locations (not migratory flights). Location data was downloaded from tags 
using PinPoint Host Software (Lotek Wireless Inc., Version 2.12.3.4).  
Migratory Connectivity Indices  
Two connectivity metrics were used to assess migratory connectivity. We calculated 
the Mantel correlation (rM) using the ade4 package in R (Dray and Dufour 2007, 
Besag et al. 2016). Pairwise orthodromic distances between breeding and winter or 
stopover sites were calculated using the geosphere package in R (Hijmans 2017). We 
estimate P-values based on 9,999 random permutations of the sets of sites.  The 
second index was the MC metric from the R package MigConnectivity (Cohen et al. 
2018). For the MC metric, breeding regions were assigned based on Rushing et al. 
(2016) delineation of natural breeding populations for wood thrush; winter and 
stopover regions followed Stanley et al. (2015) assignments.  The distance matrix was 
generated using the same methods as above. Relative breeding abundance for each 
population was derived from Rushing et al. (2016) population level estimates of 
abundance based on standardized survey routes (Breeding Bird Surveys). For both 
connectivity indices, values close to 1 indicate birds close together during the 
breeding season remain close together in the subsequent season (winter or stopover 
location). This suggests high migratory connectivity and low population mixing. 
Value close to 0 indicate low migratory connectivity and high population mixing. 
Model Specifications and Fit Assessment  
For Bayesian mixed conditional logistic regression we assumed individual level 
coefficients for all predictors were normally distributed. Vague gaussian priors were 
105 
 
 
  
used with for all population level coefficients (?i  ~ N(0,0.01)). Three separate 
Markov Chain Monte Carlo (MCMC) were run for each model with 200,000 
iterations, 100,000 burn-in, 50 thinning. Convergence was assessed using Brooks?
Gelman?Rubin statistic (rhat < 1.1; Gelman et al. 2004) and by visual inspection of 
trace plots of population and individual level coefficients. To assess model fit a 
Bayesian P-value approach was employed (Gelman et al. 1996, Valente and Betts 
2018). At each iteration in the chain (MCMC) a discrepancy statistic (D) was 
calculated as the sum of square differences between each observed data point and the 
expected value based on the fitted model. D was compared to a discrepancy statistic 
(Dsim) calculated in the same method based on a simulated data set (drawn from the 
posterior distribution of the fitted model). The p-value represented the proportion of 
time D > Dsim. Extreme Bayesian p-values (0.05 < p-value > 0.95) indicate the model 
is inadequate.  Models were ranked based on deviance information criterion (DIC) 
and we assumed models > 5 ??????? had substantial differences in model fit (Thomas et 
al. 2006).  The mean population level parameter estimates and their 95% credible 
intervals were inspected and those that did not overlap zero were interpreted as being 
important predictors in habitat selection models. All predictor variables were centered 
and standardized to improve convergence. 
To identify the top models for generalized linear and mixed effect models, 
candidate models included a fully specified global model, a null model and all 
biologically relevant combination of factors.  Competing models were evaluated 
using AICc in the R package MUMln (Barton 2017). All models < 2 ?AIC were 
considered competitive and 95% confidence were inspected. If confidence intervals 
106 
 
 
  
of additional parameters in competing models included zero and did not change the 
log-likelihood estimates they were discarded (Arnold 2010).  If more than one model 
was supported we performed unconditional model averaging of estimates and 95% 
confidence intervals in the package MUMIn (Barton 2017). Estimated marginal 
means and estimated trends were calculated using the R package emmeans (Lenth 
2018). All models were assessed for violation of model assumptions by visual 
inspection of plots of residual versus predicted values and quantile-quantile plots. 
 
  
 
107 
 
 
  
Supplementary Tables 
 
 
 
 
 
108 
 
 
Table S1.1 Number of geolocators deployed and retrieved by deployment location. Recapture rates of 
individual that did not receive GPS tags provided for comparison. 
Location Coordinates Year 
Tag 
Deploy. Retrieval 
Recapture 
Ratea Local Partners Rate 
Delaware 39.7?N, -75.7?W 2014 5/17 n/ab University of Delaware 
Indiana 39.9?N, -86.7?W 2014 5/24 7/65 (7/43) Department of Defense 
2015 2/20 9/45(7/33) 
New York 41.3?N, -73.6?W 2014 0/5 2/4 (1/3) Bedford Audubon Society 
 41.3?N, -73.7?W 2014 2c/12 6/28 (5/17) 
41.3?N, -73.7?W 2014 1/5 2/4 (2/4) 
North Carolina 36.2?N,  -80.3?W 2014 0/5 1/8 (1/4) Forsyth Audubon Society 
 36.3?N,  -80.5?W 2014 2/16 2/18 (2/12) 
Minnesota 45.1?N, -92.8?W 2014 6/18 4/14 (1/4) National Park Service 
 44.7?N, -93.0W 2014 1/7 0/2 (0/2) 
aMales in bracket 
bBirds were target netted and therefore few individuals did not receive GPS tags 
cOne additional bird re-sighted but tag not retrieved 
 
  
Table S1.2 Samples statistics for habitat use and discrete choice models used to 
examine patterns of habitat use and selection of wood thrush across the annual cycle. 
Percentage of missing location estimates not attributed to lost antenna. 
%  # Points/Bird Discrete Choice Models 
Missing 
Season Locations # Birds n ? SD Range n ? SD Range 
Breeding 30.0   20* 9.2 ? 4.2 5 - 22 5 ? 0 5 - 5 
Fall Migration 46.7 20 4.7 ? 1.7 1 ? 7 4.7 ? 1.7 1 - 7 
Winter 18.0 19 11.3 ? 3.9 2 ? 17 4.8 ? 0.7 2 - 5 
Spring Migration 49.0 16 3.8 ? 2.3 1 ? 8 3.8 ? 2.3 1 -8 
Notes: * One additional bird had a single location fix on the breeding ground for a total of 
21 individuals but was omitted from analyses. 
 
 
109 
 
 
  
. 
 
 
110 
 
 
Table S1.3 Source of local and landscape scale habitat features derived for habitat use and discrete choice models of habitat 
seHleabcittiaotn F oefa twuroeos d thrush acroUssn itth e anSncuaalel  cyDclaet.a Aseltl/ Pgreoodsupcat tial data analysis perforDmeesdc riinp tRio n3 .4.3 (R Core Team 2017) 
uLsioncga lt-hleev Rel  p ackage raster (Hijmans 201 6)   
Enhanced Vegetation n/a 250-m MODIS Land Terra Vegetation Inverse-distance-weighted interpolation at 
Index Indices 16-Day L3 Global 250ma point-estimate of pixels representing EVI 
Tree Cover % area 30-m Landsat 2012 Tree Cover Continuous Bilinear interpolation at point-estimate of pixels 
Fieldsb representing % tree cover. 
Distance to Non-Forest m 30-m Landsat 2012 Tree Cover Continuous Proximity to nearest pixel of non-forest (< 30% 
Fieldsb area of tree cover) 
Patch Size ha 30-m Landsat 2012 Tree Cover Continuous Area of horizontally or vertically connected 
Fieldsb pixels of forest (> 30% area of tree cover) where 
point-estimate located. If point-estimate found 
in non-forest value is 0. 
Landscape-level     
 Average Tree Cover % area 30-m Landsat 2012 Tree Cover Continuous Average value of % tree cover pixels located 
Fieldsb within a 1-km buffer of point-estimate 
Average Patch Size ha 30-m Landsat 2012 Tree Cover Continuous Average area of all forest patches (see above) 
Fieldsb located within a 1-km buffer of point estimate.  
Forest Loss ha 30-m Landsat 2000 ? 2014 Forest Cover Average of forest loss pixels located within a 1-
Changec,d km buffer of point-estimate. 
Land Cover Diversity n/a 0.083? MODIS Land Cover: 2001 ? 2012e Calculated Shannon Index of Diversity based on 
the proportion of land cover classes within a 5-
km buffer of point-estimate. 
a Retrieved from Env-DATA Track Annotation Service (Dodge et al. 2013). 
b Retrieved from Global Land Cover Facility (www.landcover.org; Sexton et al. 2013). 
c Retrieved from Global Land Cover Facility (Sexton et al. 2013). 
d Data downloaded using R package gfcanalysis (Zvoleff 2015). 
e Retrieved from Global Land Cover Facility (www.landcover.org; Friedl et al. 2010, Channan et al. 2014) 
 
  
Table S1.4 Candidate model sets to assess alternative hypotheses for seasonal habitat 
selection from individually tracked wood thrush. Model set run separately for each 
season (fall, spring, breeding and winter). All models included individual as a random 
factor, with the exception of the null model, to account for difference in selection 
patterns across birds. 
 
 Hypothesis Model 
Null The null model assumed probability n/a 
of use equal to random expectations 
(1 / # choice sets). If consistently 
chosen it would suggest geospatial 
data not adequate to predict habitat 
selection. 
Local The local model would be % Tree Cover + EVI + Patch Size 
appropriate if resources units were + Proximity to Non-Forest  
selected based on local habitat 
features only.  
Landscape1 The landscape model would be % Tree Cover 1Km + Average 
appropriate if resources units were Patch Size 1km 
selected based on landscape habitat 
features.  
Full1 The full model would be appropriate % Tree Cover + EVI + Patch Size 
if resource units were selected for + Proximity to Non-Forest +% 
both local and landscape habitat Tree Cover 1Km + Average Patch 
features  Size 1km 
1Only run for 15 km and 25 km spatial extent 
 
 
111 
 
 
  
Table S1.5 Delta deviance information criteria (?DIC) model selection for discrete 
choice models that examined habitat selection of male wood thrushes from 5 breeding 
population across the annual cycle. For each spatial extent and season, a candidate set 
of models were run, the parameters included in each model found in Table S1.4. All 
models < 5 ?DIC were considered competitive. 
 
Spatial Extent 
Season 0.5 km 5 km 15 km 25 km 
Breeding     
Null 219.26 336.29 295.51 109.47 
Patch 0.00 0.00 0.00 101.58 
Landscape n/a n/a 227.78 48.37 
Patch + Landscape n/a n/a *4.86 0.00 
Fall     
Null 185.19 195.12 80.45 65.41 
Patch 0.00 0.00 0.00 0.00 
Landscape n/a n/a 82.78 63.66 
Patch + Landscape n/a n/a *3.12 6.25 
Spring     
Null 0.00 0.00 0.00 0.00 
Patch 23.15 14.48 13.46 72.14 
Landscape n/a n/a *4.41 6.59 
Patch + Landscape n/a n/a 19.36 70.29 
Winter     
Null 157.48 205.43 130.52 103.22 
Patch 0.00 0.00 0.00 0.00 
Landscape n/a n/a 104.02 70.17 
Patch + Landscape n/a n/a *1.23 9.16 
Notes: Bolded values indicate top models. Asterix indicate 
models with ?DIC < 5, which have the same parameter 
estimates with 95% credible intervals that do not cross zero as 
the top model. 
112 
 
 
  
Table S1.6 AICc model selection for generalized linear models of seasonal and 
population differences in habitat features for male wood thrush from 5 breeding 
populations. All models included random intercept for individuals and models < 2 
?AICc were considered competitive.  
 
Model n Variables df Log- AICc ?AICc 
likelihood 
Patch Level       
Tree Cover 510 Season  6 -1977.47 3967.11 0.00 
(%)  Breed. Long. + Season 7 -1978.16 3970.55 3.43 
  Breed. Long. * Season 10 -1975.85 3972.15 5.03 
  Null 3 -2014.45 4034.95 67.84 
  Breed. Long. 4 -2015.08 4038.24 71.13 
EVI 508 Season  6 506.16 -1000.14 0.00 
  Breed. Long. + Season 7 502.17 -990.11 10.04 
  Null 3 490.28 -974.51 25.63 
  Breed. Long. 4 485.64 -963.19 36.95 
  Breed. Long. * Season 10 488.48 -956.52 43.62 
Patch Size (ha) 549 Breed. Long. * Season 10 -8122.12 16264.65 0.00 
  Breed. Long. + Season 7 -8156.41 16327.02 62.37 
  Season 6 -8168.41 16348.98 84.33 
  Breed. Long. 4 -8357.81 16723.69 459.04 
  Null 3 -8369.47 16744.99 480.34 
Prox. to Non- 493 Breed. Long. * Season 10 -959.64 1939.73 0.00 
Forest (m)  Season  6 -990.66 1993.50 53.77 
  Breed. Long. + Season 7 -991.50 1997.24 57.51 
  Null 3 -1030.54 2067.13 127.41 
  Breed. Long. 4 -1031.45 2070.98 131.25 
Landscape       
level 
Avg. Tree 540 Breed. Long. * Season 10 -2084.52 4189.46 0.00 
Cover 1 km  Season  6 -2095.97 4204.09 14.63 
(%)  Breed. Long. + Season 7 -2096.48 4207.18 17.72 
  Null 3 -2228.84 4463.73 274.27 
  Breed. Long. 4 -2229.33 4466.74 277.28 
Avg. Patch 540 Breed. Long. * Season 10 -7949.99 15920.40 0.00 
Size 1 km (ha)  Breed. Long. + Season 7 -7993.79 16001.78 81.38 
  Season 6 -8004.39 16020.93 100.53 
  Breed. Long. 4 -8054.86 16117.79 197.39 
  Null 3 -8065.45 16136.95 216.54 
Land Cover      
Land Cover  Season 6 -26.15 65.54 0.00 
5km  Breed. Long. + Season 7 -30.41 76.50 10.96 
  Breed. Long. * Season 10 -39.35 102.13 36.60 
  Null 3 -55.71 117.76 52.23 
113 
 
 
  
: Seasonal Changes in Habitat Features 
Supplement to Chapter 1 
 
To better interpret habitat selection patterns (i.e. the behavioural decision making 
process) we investigated how habitat-use patterns changed across the annual cycle 
(Jones 2001). We paired location fixes to geospatial habitat data (detailed description 
of methods in Appendix S1, Table S1.3) to determine how habitat features (local and 
landscape) of locations occupied by wood thrush changed across the annual cycle. A 
generalized linear model was performed for each habitat characteristic (local and 
landscape), the full model included the habitat variable as the response variable, the 
interaction of season and breeding longitude as fixed effects and individual as a 
random effect. The best-fit model for each habitat features was used to calculate the 
repeatability in use of habitat features. 
The characteristics of habitats occupied by wood thrush varied significantly 
between seasons for all local and landscape habitat features. Marginal means of most 
habitat features (local and landscape) were highest during the winter period (Table 
S2.1), indicative of more forested and productive habitats (high EVI). Most habitat 
features had the lowest marginal mean values during the migratory periods and there 
was no significant difference in marginal mean values between migratory periods 
(Table S2.1). 
114 
 
 
  
 
 
 
115 
 
 
Table S2.1 Estimated marginal means and 95% confidence intervals of the effect of season on local and landscape level 
(1 km radius) environmental characteristics of habitats occupied by wood thrush across the annual cycle. 
 
 Patch-level Characteristics 
 Tree Cover (%) EVI Patch Size (km2) Prox. to Non-Forest (m) 
Season Means 95% CI Means 95% CI Means 95% CI Means 95% CI 
         
Breeding 56.2a 53.1 ? 59.3 0.52a 0.52 ? 0.54 *1972a 0 ? 4220 *99.2a 57.8 ? 170.3 
Fall Migration 55.3ab 51.7 ? 58.8 0.46b 0.46 ? 0.48 8738b 6025.5 ? 11270.4 *19.7c 10.8 ? 35.8 
Winter 63.5c 60.4 ? 66.7 0.53c 0.53 ? 0.55 17714c 15498.9 ? 19929.2 *38.0b 21.8 ? 66.4 
Spring Migration 50.4b 46.1 ? 54.7 0.48ab 0.48 ? 0.51 6141ab 3305.5? 8977.1 10.2c 5.2 ? 19.9 
 Landscape-level Characteristics  
 Tree Cover 1km (%) Avg. Patch Size (km2)    
Season Means 95% CI Means 95% CI     
Breeding 39.5a 35.5 ? 43.5 33045a 30876.3 ? 35233.9     
Fall Migration 48.1b 43.7 ? 52.4 29123b 26697.2 ? 31548.8     
Winter 60.3c 56.3 ? 64.2 *28489b 26346.9 ? 30631.7     
Spring Migration 43.2ab 38.3 ? 48.1 27378b 24583.2 ? 30174.2     
Notes: Letters denote values that are significantly different from each other using Tukey?s adjustments for multiple comparisons. Asterix 
indicates a significant interaction with breeding longitude. 
 
  
: Population-specific annual schedules 
Supplement to Chapter 1 
 
To examine the consistency of annual schedules between breeding populations, the 
timing of annual events was determined for all individuals. Due to low success of 
location fixes during migratory periods, annual cycle stages were defined based on 
the first or last location recorded at stationary stages (i.e. spring migration ends day 
before first recorded point on breeding grounds). Migratory periods were identified as 
northward or southward movements over the course of consecutive days (+2 points, 6 
- 8 days) and larger than 20km. To determine if population-specific annual schedule 
were present we examined the timing of migratory events across breeding populations 
using generalized linear model. The full model included, timing event (start, end, 
duration, rate, day of gulf cross) as the response variable and breeding latitude, 
breeding longitude and winter longitude as fixed effects. Winter arrival dates could 
not be reliably assigned for some individuals due to low frequency of location fixes 
(bimonthly fixes) at arrival. Therefore, analysis of the fall migration period was 
restricted to departure from breeding grounds to departure from Gulf of Mexico. 
We found evidence that timing of both spring and fall migration were 
influenced by breeding location, but not winter location (Figure S3.1; Table S3.1). 
Breeding latitude and longitude had a negative effect on fall migration departure, 
more northern and eastern populations departed earlier on fall migration (Table S3.2). 
There was no evidence that latitude or longitude of breeding and winter location 
influenced date of arrival at the Gulf of Mexico during fall migration or the number 
of days on route to the Gulf of Mexico (mean ? SE = 26.7 ? 10.9 days). During 
116 
 
 
  
spring migration, winter departure date, arrival at the Gulf of Mexico and breeding 
arrival date were positively predicted by breeding latitude, indicating southern 
breeding populations had an earlier migration schedule than northern populations 
(Table S3.2). There was no evidence that breeding longitude or winter location 
influenced spring migration departure or arrival dates (Table S3.2).  Spring migration 
duration (mean ? SE = 12.2 ? 5.6 days) and speed were not influenced by breeding 
and winter location (Table S3.2). 
 
 
117 
 
 
  
Table S3.1AICc model selection results for generalized linear models evaluating the 
influence of breeding and winter location on migration timing events of male wood 
thrush from 5 breeding populations. All models < 2 ?AICc were considered 
competitive. 
 n Variables df Log- AICc ?AICc 
Model likelihood 
Fall Migration     
Breeding 20 Breed. Lat. + Breed. Long. 4 -62.18 135.22 0.00 
Departure  Breed. Lat. + Breed. Long. + W Lat. 5 -62.00 138.62 3.40 
  Null 2 -67.54 139.84 4.62 
  Breed. Lat. 3 -66.43 140.47 5.25 
Gulf 13 Null 2 -38.50 82.33 0.00 
Coast  W Lat.  
Arrival 3 -38.07 85.14 2.80 
  Breed. Long. + W Lat. 4 -37.85 89.42 7.09 
  Breed. Lat. + Breed. Long. + W Lat. 5 -36.70 93.39 11.06 
Duration 13 Null 
to Gulf 2 -43.84 93.00 0.00 
  W Lat.  3 -42.31 93.62 0.61 
  Breed. Long. + W Lat. 4 -42.29 98.28 5.28 
  Breed. Lat. + Breed. Long. + W Lat. 5 -42.24 104.48 11.48 
Spring Migration     
Winter 16 Breed. Lat. 3 -48.84 105.69 0.00 
Departure  Breed. Lat. + Breed. Long. 4 -48.57 108.77 3.08 
  Breed. Lat. + Breed. Long. + W Lat. 5 -48.01 112.01 6.33 
  Null 2 -53.57 112.07 6.38 
Gulf 12 Breed. Lat. 3 -35.10 79.20 0.00 
Coast  Breed. Lat. + Breed. Long. 4 -34.93 83.57 4.37 
Arrival  Null 2 -41.04 87.41 8.21 
  Breed. Lat. + Breed. Long. + W Lat. 5 -34.89 89.78 10.58 
Breeding 14 Breed. Lat. + Breed. Long. 4 -38.91 90.26 0.00 
Arrival  Breed. Lat. 3 -41.74 91.88 1.62 
  Breed. Lat. + Breed. Long. + W Lat. 5 -38.86 95.22 4.96 
  Null 2 -49.20 103.49 13.23 
Duration 14 W Lat. 3 -40.99 90.38 0.00 
  Null 2 -43.41 91.91 1.52 
  Breed. Lat. + W Lat. 4 -40.00 92.44 2.05 
  Breed. Lat. + Breed. Long. + W Lat. 5 -39.57 96.64 6.26 
Speed 14 Null 2 -67.94 140.98 0.00 
  W Lat. 3 -67.11 142.61 1.63 
  Breed. Long. + W Lat. 4 -66.60 145.64 4.67 
  Breed. Lat. + Breed. Long. + W Lat. 5 -66.43 150.37 9.39 
 
 
118 
 
 
  
Table S3.2 Model averaged parameter estimates and unconditional 95% confidence 
intervals for top linear models examining the influence of breeding location (latitude 
and longitude) and winter location (latitude and longitude) on migration timing events 
of male wood thrush from 5 breeding populations. 
 
Model Variable Estimate 95% CI 
Fall Migration    
Breeding Departure Intercept 5.70 5.53 ? 5.88 
 Breeding latitude -0.0094 -0.004 - -0.0044 
 Breeding longitude -0.0032 -.0052 - -0.0013 
    
Gulf Coast Arrival Null   
    
Duration to Gulf Null   
    
Spring Migration    
Winter Departure Intercept 3.99 3.62 ? 4.36 
 Breeding latitude 0.015 0.0064 ? 0.024 
    
Gulf Coast Arrival Intercept 29.50 -7.43 ? 66.44 
 Breeding latitude 1.86 0.97 ? 2.75 
    
Breeding Arrival Intercept 37.92 -0.49 ? 76.34 
 Breeding latitude 2.80 1.39 ? 4.21 
 Breeding longitude 0.36 -0.60 ? 0.95 
    
Duration Null   
    
Speed Null   
119 
 
 
  
Figure S3.1 Spatiotemproal organization of the annual cycle of five breeding 
populations of wood thrush tracked using GPS tags (n = 20). Each horizontal bar 
represents an individual and colours represent different stages of the annual cycle. 
Dashed vertical lines represent mean start and end date of fall and spring migration, 
respectively.            
 
120 
 
 
  
 
Supplement to Chapter 2: 
 Supplementary Methods 
 Supplementary Tables S1-S7 
Supplementary Figure S1 
 
Supplementary Methods 
Genetic Sexing 
Genomic DNA was extracted from whole blood (~50 ?l) stored in Queen?s lysis 
buffer using the BioSprint 96 Tissue DNA Kit (Qiagen, Valencia, CA).  Sex was 
determined by amplifying genomic DNA (2 ?l) using highly conserved P2 and P8 
primers (Griffiths et al. 1998) in a 11 ?l polymerase chain reaction. PCR products 
were size-sorted on an Applied Biosystems 3130 Genetic Analyzer (ABI, Foster City, 
CA) run with GeneScan 500 base pair ROX internal size standard (ABI). Samples 
were scored using Genemapper v.4.1 (ABI). Both sexes carry a CHD-Z gene (~380 
bp long), only the females carry the CHD-W gene (~400 bp long).  
Home Range Analysis 
To derive utilization distributions, location data for each individual was first fit to 
three continuous-time movement models using the ctmm package in R (Calabrese et 
al. 2016). The first model was considered a null model assuming independent and 
identically distributed data (IID, equivalent to traditional kernel density estimators). 
The second model fit locations to a Brownian motion model with restricted space-use 
(Ornstein-Uhlenbeck). The final model represented an extension of the second model 
and introduced both correlated velocities and restricted space-use (Ornstein-
Uhlenbeck Foraging). Models were ranked using the Akaike?s Information Criterion 
121 
 
 
  
corrected for small sample size (Burnham and Anderson 2002) and the top model was 
used to fit the Autocorrelated Kernel Density Estimate. The best-fit model for the 
majority of individuals (n = 31) was the IID, followed by the Ornstein-Uhlenbeck 
Foraging (n = 13) and Ornstein-Uhlenbeck model (n = 9). 
Stable-isotope analysis 
Whole blood samples were kept frozen until they were oven dried at ~100? for 24 
hours. Blood was powdered and approximately 0.5 mg weighed and packaged in a tin 
capsule. Claw samples were cleaned in 2:1 choloroform:methanol solution and air 
dried in a fume hood for 24 hours. Approximately 0.5g of claw was cut from the 
distal end of the claw, weighed and packaged in tin capsules. All samples were 
analyzed at the Smithsonian OUSS/MCI Stable Isotope Mass Spectrometry 
Laboratory (Suitland, MD).  Samples were crushed and combusted on a Costech 
Elemental Analyzer and introduced to a Thermo Delta V Advantage mass 
spectrometer through a Thermo Conflo IV. For every 10-12 samples a set of 
standards were run that included Costech Acetanilide and Urea-UIN31 calibrated to 
USGS40 (L-glutamic acid) and USGS41 (L-glutamic acid). Isotope values were 
corrected using a 2-point linear correction on the calibrated standards. ?13C and 
?15N values were reproducible to ?0.2?. 
Remote Sensing 
To assess pattern of moisture at our study site in Belize and for GPS tracked 
individuals we derived the wetness metric based on the tasseled-cap transformation of 
30-m resolution Landsat imagery (Crist and Kauth 1986). At our study site in Belize 
we derived the wetness metric for each plot and home range and for GPS-tracked 
122 
 
 
  
individuals we derived the wetness metric for each relocation. A plot level index of 
wetness was derived in order to perform analyses that included individuals that 
departed our study site before sufficient relocations were accumulated to estimate 
home range area. Landsat images are available on a bi-monthly basis from locations 
across the globe. Level-1 Landsat images were obtained from our study areas in 
Belize from December to April 2013 to 2015 and for GPS-tracked birds from 
December to April 2014-2016 from the USGS database EarthExplorer (Landsat 7 
+ETM in 2013, Landsat 8 OLI from 2014 to 2016). Differences in spectral responses 
between Landsat sensors have been found, however comparisons of vegetation 
indices derived from across Landsat sensors suggest a high continuity across sensor 
types, particularly for band-transformation derived vegetation indices such as the 
tasseled cap transformation (Li et al. 2013, She et al. 2015).  
All image processing was performed in R. In brief, digital numbers were 
converted to top-of-atmosphere reflectance using the radCor function in RStoolbox 
(cit) and the Quality Assessment band was used to create a cloud mask. The 
tasseledCap function in RStoolbox was used to derive the wetness metric. In Belize, 
for each image, we extracted the mean wetness metric across each (i) individual home 
range and (ii) plot, as defined by a 1-km radius at the centroid of tracking locations 
using the extract function in the R package raster (cite). We excluded values derived 
from home ranges that were masked > 90% by contamination (over 50% of values 
were excluded). For GPS transmitters, we extracted mean wetness from each point 
using bilinear interpolation in the raster package. For birds that made intra-winter 
movements, only locations prior to the first movement were included. In some 
123 
 
 
  
instances, multiple tiles covered locations, to account for this we took the average 
wetness value for every two-week period.   
124 
 
 
  
Supplementary Tables 
 
 
 
  
125 
 
 
Table S4.1 Summary statistics for population and environmental condition variables collected 
from non-breeding wood thrush (a) radio tracked in Belize from 2013 to 2015 and (b) 
monitored with GPS tags across the non-breeding range. 
Mean (? SE) 
Variables Year Average/ Overall 
2013 2014 2015  
(a) Belize Radio tracking     
Population     
Age Ratio (Adult:Yearling) 36:30 24:23 29:16 86:69 
Sex Ratio (Male:Female:Unk) 23:29:14 15:30:2 23:20:2 58:72:17 
Body Condition Index 2.8 ? 0.64 3.7 ? 0.83 1.7 ? 0.97 2.7 ? 0.48 
Home Range Size (ha) 1.6 ? 0.20 1.6 ? 0.45 3.4 ? 0.76 2.3 ? 0.37 
Departure Dates (Julian date) 79 ? 2.5 67 ? 2.7 65 ? 3.0 72 ? 1.6 
     
Environmental Condition     
Fruit Counts 39.5 ? 19.97 104.1 ? 21.80 51.0 ? 7.89 55.4 ? 10.46 
Insect Counts 8.72 ? 0.77 14.9 ? 0.70 16.6 ? 0.62 12.6 ? 0.50 
?15N Claws 9.3 ? 0.15 8.6 ? 0.15 8.8 ? 0.16 8.9 ? 0.09 
?15N Blood 10.3 ? 0.12 9.7 ? 0.16 9.8 ? 0.16 9.9 ? 0.09 
Home range Wetness -0.06 ? 0.002 0.04 ? 0.001 0.04 ? 0.0008 0.01? 0.006 
Plot Wetness -0.06 ? 0.001 0.04 ? 0.003 0.04 ? 0.002 0.008 ? 0.0219 
?13C Claws -23.0 ? 0.06 -23.4 ? 0.04 -23.1 ? 0.05 -23.2 ? 0.03 
?13C Blood -24.9 ? 0.09 -25.1 ? 0.06 -24.9 ? 0.06 -25.0 ? 0.04 
     
(b) GPS tracking     
Departure Dates (Julian Week)    15 ? 0.5 
Wetness    0.03 ? 0.003 
 
  
Table S4.2 AICc model selection for generalized mixed effect models describing 
home range size (95% isopleth of utilization distribution) of non-breeding wood 
thrush in Belize.  Age (yearling or adult), sex (male or female), stable-carbon values 
derived from claws or blood and mean satellite-derived wetness, fruit counts and 
insect counts for each home range were included as fixed effects. All models included 
a random intercept for year and models < 2 ?AICc were considered competitive. 
Separate models were constructed for isotope values derived from claws (n = 37) 
which provides information on diets over a few months and blood (n = 44) which 
provides information on diets over a few weeks. 
 
Model df Log- AICc ?AI
likelihood Cc 
Claw     
Age + Sex 5 -58.90 129.74 0.00 
Age * Sex 6 -58.57 131.93 2.19 
Age + Sex + ?13C + Condition 7 -57.34 132.54 2.80 
Age + Sex + FruitCounts + InsectCounts 7 -57.63 133.13 3.38 
Age + Sex + Wetness + Condition 7 -57.97 133.81 4.07 
Age + Sex + ?13C + ?15N 7 -58.02 133.91 4.17 
Age + Sex + FruitCounts + InsectCounts + Condition 8 -57.56 136.26 6.51 
FruitCounts + InsectCounts 5 -64.49 140.93 11.18 
Null 3 -67.41 141.54 11.79 
Wetness 4 -67.14 143.53 13.79 
Condition 4 -67.26 143.77 14.03 
Wetness + Condition 5 -67.04 146.02 16.28 
?13C + Condition 5 -67.24 146.42 16.68 
     
Blood     
Age + Sex + ?13C + Condition 7 -73.85 164.81 0.00 
?13C + Condition 5 -76.92 165.42 0.62 
Age + Sex + ?13C + ?15N 7 -74.45 166.00 1.20 
Age + Sex 5 -77.51 166.60 1.79 
Age * Sex 6 -76.84 167.96 3.15 
FruitCounts + InsectCounts 5 -78.28 168.14 3.33 
Null 3 -81.00 168.60 3.79 
Age + Sex + FruitCounts + InsectCounts 7 -76.32 169.74 4.94 
Wetness 4 -80.78 170.59 5.78 
Condition 4 -81.00 171.02 6.21 
Age + Sex + Wetness + Condition 7 -77.35 171.81 7.01 
Age + Sex + FruitCounts + InsectCounts + Condition 8 -76.23 172.57 7.76 
Wetness + Condition 5 -80.77 173.12 8.31 
 
126 
 
 
  
Table S4.3 AICc model selection for general linear models evaluating variation in 
fruit (n = 44) and insect (n = 43) counts conducted on non-breeding home ranges of 
wood thrush during the early dry season sampling period (mid-February) in Belize.  
Age (yearling or adult), sex (male or female), mean satellite-derived wetness of home 
range and home range size (area of 95% isopleth of utilization distribution) were 
included as fixed effects. Models < 2 ?AICc were considered competitive.  
 
Model df Log- AICc ?AICc 
likelihood 
Fruit Counts     
Wetness 3 -50.34 107.28 0.00 
Age * Sex + Wetness + Area 7 -46.43 109.96 2.68 
Age + Sex + Wetness 5 -49.82 111.21 3.93 
Age + Sex + Wetness + Area 6 -49.32 112.91 5.63 
Null 2 -61.93 128.15 20.87 
Area 3 -61.86 130.32 23.04 
Age + Sex 4 -60.78 130.58 23.30 
Age * Sex 5 -59.74 131.07 23.79 
Age + Sex + Area 5 -60.75 133.07 25.79 
     
Insect Counts     
Wetness 3 -53.44 113.49 0.00 
Age + Sex + Wetness 5 -52.54 116.70 3.21 
Age + Sex + Wetness + Area 6 -51.66 117.65 4.15 
Age * Sex + Wetness + Area 7 -51.56 120.31 6.82 
Null 2 -60.51 125.32 11.83 
Area 3 -60.43 127.47 13.98 
Age + Sex 4 -60.07 129.19 15.70 
Age + Sex + Area 5 -59.86 131.35 17.86 
Age * Sex 5 -60.06 131.74 18.25 
 
 
 
127 
 
 
  
Table S4.4 AICc model selection for linear mixed effect models evaluating variation 
in ?15N values in the blood or claws of non-breeding wood thrush in Belize across 
the dry season. ?15N values provide a proxy for proportion of insects consumed.  Age 
(yearling or adult), sex (male or female), mean satellite-derived wetness of plots, 
?13C (proxy for habitat moisture) and sampling period (early or late) were included 
as fixed effects. All models included a random intercept for bird identity nested 
within year and models < 2 ?AICc were considered competitive. Separate models 
were constructed for isotope values derived from claws (n = 148) which provides 
information on diets over a few months and blood (n = 165) which provides 
information on diets over a few weeks. 
 
Model df Log- AICc ?AICc 
likelihood 
Claws     
Wetness * Period 6 -185.30 383.20 0.00 
Wetness + Period 5 -186.93 384.28 1.08 
Age + Sex + Wetness + Period 7 -186.21 387.22 4.02 
Wetness + ?13C + Period 6 -187.43 387.46 4.26 
Age * Sex + Wetness + Period 8 -186.18 389.39 6.19 
Age + Sex + Wetness * Period + ?13C 9 -185.12 389.55 6.35 
Age + Sex + Wetness + ?13C + Period 8 -186.74 390.52 7.32 
Null 3 -190.85 392.12 8.92 
?13C + Period 5 -186.68 392.66 9.46 
Age * Sex + Wetness + ?13C + Period 9 -189.35 393.49 10.29 
Age + Sex + ?13C + Period 7 -192.45 397.49 14.29 
     
Blood     
Wetness * Period 6 -213.87 440.27 0.00 
Wetness + Period 5 -215.92 442.21 1.94 
Wetness + ?13C + Period 6 -214.95 442.44 2.16 
?13C + Period  5 -217.56 445.49 5.22 
Age + Sex + Wetness * Period + ?13C  9 -214.03 447.22 6.95 
Age + Sex + Wetness + Period 7 -216.94 448.59 8.32 
Age + Sex + Wetness + ?13C + Period 8 -215.96 448.83 8.56 
Age + Sex + ?13C + Period 7 -218.27 451.25 10.98 
Age * Sex + Wetness + Period 8 -217.17 451.26 10.98 
Age * Sex + Wetness + ?13C + Period 9 -216.20 451.55 11.28 
Null 3 -222.74 451.62 11.35 
 
 
 
128 
 
 
  
Table S4.5 AICc model selection for general linear mixed effect models evaluating 
variation in the body condition index of non-breeding wood thrush during the early 
dry season sampling period (mid-January ? mid-February) in Belize.  Age (yearling 
or adult), sex (male or female), mean satellite-derived wetness of plots and ?13C 
(proxy for habitat moisture) and ?15N (proxy for proportion of insects) values were 
included as fixed effects. Separate models were run including isotope values derived 
from either claws (n = 104) which provides information on diets over a few months 
and blood (n = 112) which provides information on diets over a few weeks.  Year was 
included as a random effect and models < 2 ?AICc were considered competitive. 
 
Model df Log- AICc ?AICc 
likelihood 
Claw     
Age * Sex + Wetness * ?13C 9 -316.41 652.74 0.00 
Age * Sex + Wetness * ?15N 9 -317.29 654.50 1.76 
Age * Sex + Wetness + ?13C + ?15N 9 -320.25 660.41 7.67 
Age * Sex + Wetness + ?15N 8 -321.74 660.99 8.25 
Age * Sex + Wetness + ?13C 8 -322.52 662.55 9.81 
Age * Sex 6 -327.64 668.14 15.40 
Age * Sex + ?13C 7 -326.63 668.43 15.70 
Null 3 -335.59 677.42 24.69 
     
Blood     
Age * Sex + Wetness + ?13C + ?15N 9 -339.57 698.90 0.00 
Age * Sex + Wetness * ?15N 9 -339.79 699.35 0.45 
Age * Sex + Wetness * ?13C 9 -340.38 700.52 1.62 
Age * Sex + Wetness + ?15N 8 -345.31 708.03 9.12 
Age * Sex + Wetness + ?13C 8 -345.67 708.73 9.83 
Age * Sex + ?13C 7 -350.05 715.18 16.28 
Age * Sex 6 -352.93 718.66 19.76 
Null 3 -360.15 726.53 27.63 
 
129 
 
 
  
Table S4.6 AICc model selection for general linear mixed effect models evaluating 
differences in weight change of non-breeding wood thrush across our early (mid-
January ? mid-February) and late sampling period (mid-March) in Belize.  Age 
(yearling or adult), sex (male or female), mean satellite-derived wetness of plots and 
?13C (proxy for habitat moisture) and ?15N (proxy for proportion of insects) value 
were included as fixed effects. Separate models were run including isotope values 
derived from either claws (n = 26) which provides information on diets over a few 
months and blood (n = 31) which provides information on diets over a few weeks.  
Year was included as a random effect and models < 2 ?AICc were considered 
competitive. 
 
Model df Log- AICc ?AICc 
likelihood 
Claw     
Age * Sex + Wetness + ?13C  8 -48.41 121.29 0.00 
Age * Sex + Wetness + ?15N 8 -49.16 122.79 1.50 
Null 3 -58.16 123.41 2.12 
Age * Sex 6 -53.82 124.06 2.77 
Age * Sex + Wetness + ?13C + ?15N 9 -48.19 125.62 4.33 
Age * Sex + ?13C 7 -53.01 126.24 4.95 
Age * Sex + ?15N 7 -53.48 127.19 5.90 
     
Blood     
Age * Sex + Wetness + ?13C 8 -59.57 141.69 0.00 
Age * Sex + Wetness + ?13C + ?15N 9 -57.93 142.43 0.74 
Age * Sex + Wetness + ?15N 8 -60.22 142.99 1.30 
Age * Sex + ?13C 7 -63.07 145.00 3.31 
Null 3 -69.80 146.49 4.80 
Age * Sex + ?15N 7 -64.15 147.17 5.48 
Age * Sex 6 -66.16 147.81 6.12 
 
130 
 
 
  
Table S4.7 AICc model selection for Cox proportional hazard models evaluating site 
persistence of non-breeding wood thrush in Belize.  Age (yearling or adult), sex (male 
or female), early season body condition index, mean satellite-derived wetness of 
plots, ?13C (proxy for habitat moisture) and ?15N (proxy for proportion of insects) 
isotopic signatures and year were included as fixed effects. Separate models were run 
including isotope values derived from either claws (n = 91) which provides 
information on diets over a few months and blood (n = 101) which provides 
information on diets over a few weeks.  Year was included as a random effect and 
models < 2 ?AICc were considered competitive. 
 
Model df Log- AICc ?AI
likelihood Cc 
Claw     
?13C + ?15N + Condition + Year 5 -271.90 554.72 0.00 
?13C + ?15N + Year 4 -273.70 556.00 1.28 
Age + Sex + ?13C + ?15N + Condition + Year 7 -271.23 558.21 3.49 
Age * Sex + ?13C + ?15N + Condition + Year 8 -270.92 560.13 5.41 
Age + Sex + ?13C + Wetness + Condition + Year 7 -272.32 560.38 5.66 
Age + Sex + ?13C + ?15N + Wetness + Condition + 8 -271.18 560.65 5.93 
Year 
Age + Sex + ?13C * Wetness + Condition + Year 8 -271.21 560.70 5.98 
Wetness + Year 3 -278.13 562.61 7.90 
Age * Sex + ?13C + ?15N + Wetness + Condition + 9 -270.87 562.65 7.93 
Year 
Condition + Year 3 -278.22 562.79 8.08 
Age + Sex + Wetness + Year 5 -277.70 566.32 11.60 
Null 0 -284.85 569.70 14.98 
     
Blood     
Condition + Year 3 -306.53 619.38 0.00 
Wetness + Year 3 -306.60 619.53 0.15 
?13C + ?15N + Year 4 -306.39 621.34 1.95 
?13C + ?15N + Condition + Year 5 -306.03 622.90 3.51 
Age + Sex + Wetness + Year 5 -306.41 623.66 4.28 
Null 0 -313.57 627.14 7.76 
Age + Sex + ?13C + ?15N + Condition + Year 7 -305.77 627.16 7.77 
Age + Sex + ?13C + Wetness + Condition + Year 7 -306.02 627.65 8.27 
Age + Sex + ?13C + ?15N + Wetness + Condition + 8 -305.68 629.47 10.09 
Year 
Age * Sex + ?13C + ?15N + Condition + Year 8 -305.77 629.65 10.27 
Age + Sex + ?13C * Wetness + Condition + Year 8 -306.01 630.14 10.76 
Age * Sex + ?13C + ?15N + Wetness + Condition + 9 -305.68 632.04 12.66 
Year 
 
131 
 
 
  
Table S4.8 AICc model selection for Cox proportional hazard models evaluating site 
persistence of non-breeding adult male wood thrush tracked using GPS transmitters 
from 5 breeding populations (n = 16).  Satellite-derived wetness and coefficient of 
variation of wetness values interpolated from location estimates and winter latitude 
were included as fixed effects. Models < 2 ?AICc were considered competitive. 
 
Model df Log- AICc ?AICc 
likelihood 
Wetness 1 -15.76 34.33 0.00 
Wetness + cvWetness 2 -13.96 34.92 0.59 
Null 0 - 18.40 36.80 2.47 
cvWetness 1 -17.46 37.73 3.40 
Latitude 1 -17.48 37.76 3.44 
Wetness + Latitude 2 -15.63 38.26 3.94 
cvWetness + Latitude 2 -16.77 40.55 6.22 
 
132 
 
 
  
Supplementary Figures 
 
Figure S4.1 Change in environmental (A ? C) and body condition (D) for non-
breeding wood thrush in Belize between early and late dry season sampling points. 
(A) Decrease in satellite-derived home range wetness between the early (December ? 
February) and late (March ? April) dry season. Increase in (B) fruit counts and 
decrease in (C) insect counts performed on wood thrush home ranges between early 
(mid-February) and late (end of March) sampling periods. Decrease in mass (D) of 
wood thrush capture during the early (mid-January to mid-February) and late (mid-
March) sampling periods. 
133 
 
 
  
: Relationship between stable carbon isotopes and satellite wetness 
Stable isotopes have been extensively used in ecological studies because stable 
isotope ratios in the tissues of animals reflect that of their supporting food web 
(Hobson 2011). For stable-carbon isotopes differences in the photosynthetic pathways 
and water-use efficiency in plants drive variation in stable-carbon isotopic signatures 
which can be used to identify habitat occupancy (Michener and Lajtha 2007, Kohn 
2010). C3 plants incorporate less 13C during photosynthesis compared with C4 and 
CAM plants, leading to more depleted ?13C values in their tissues. Furthermore, 
within C3 plants water-use efficiency also influences ?13C values and plant tissue 
becomes become more enriched as water availability decreases. C3 plants are the 
dominant plants in forested habitats and can therefore be used to infer an index of 
habitat moisture across the home range of individual birds (Smith et al. 2011b).  
To establish that stable-carbon values at our study site in Belize reflect 
moisture gradients across habitats we examined the factors that drove variation in 
stable-carbon values. We developed a candidate model set that reflected a priori 
hypotheses for variation in ?13C values. We hypothesized the moisture, as measured 
by satellite-derived wetness metric, would drive variation in ?13C values. We also 
hypothesized that dominance relationships between age and sex classes may influence 
habitat occupancy and therefore ?13C values. Stable-nitrogen values were included in 
the model because differences in the proportion of insects consumed has also been 
found to influence stable carbon values (Pearson et al. 2003). Finally, period was 
included in the analysis as a fixed effect because we expected moisture to decline as 
134 
 
 
  
the dry season progressed. Separate models were run to include plot wetness or home 
range wetness as a fixed effect. Bird identity nested within year as a random effect. 
 As predicted, we found evidence that stable-carbon values reflected habitat 
moisture as measured by the satellite-derived wetness metric (Figure S5.1, Table S5.1 
and Table S5.2). Wetness measured at the plot level negatively influenced stable-
carbon values in both blood (? = -5.98, 95% CI = -10.69, -1.27) and claws (? = -7.15, 
95% CI = -11.85, -2.45; Table S5.3). As plot wetness decreased, ?13C became more 
enriched (positive) in the tissues of wood thrush. The sampling period also influenced 
stable-carbon values in blood (? = -0.54, 95% CI = -0.85, -0.22). ?13C values became 
more depleted during the late sampling period. Separate models were constructed 
including individuals with known home ranges, although home range wetness was 
included in the top model with stable isotopes derived from blood and claw (Table 
S5.2) the confidence intervals crossed zero (Table S5.3). Models including home 
range wetness had a smaller sample size compared to models using plot wetness 
which may have contributed to differences in parameter estimates. Overall, these 
results suggest, that similar to other studies examining habitat occupancy in migratory 
songbirds (Marra et al. 1998, Drake et al. 2013), stable-carbon values in the tissues of 
wood thrush can be used to reflect habitat moisture.  
 
135 
 
 
  
Table S5.1 AICc model selection for generalized mixed effect models describing 
variation in stable-carbon values of non-breeding wood thrush in Belize across two 
different plots. Age (yearling or adult), sex (male or female), mean satellite-derived 
wetness of plots, ?15N (proxy for proportion of insects consumed) and sampling 
period (early or late) were included as fixed effects. Separate models were run 
including isotope values derived from either claw (n = 147) which provides 
information on diets over a few months and blood (n = 164) which provides 
information on diets over a few weeks. All models included a random intercept for 
bird identity nested within year and models < 2 ?AICc were considered competitive. 
 
Model df Log- AICc ?AICc 
likelihood 
Claw     
Wetness * Period 6 -191.90 396.39 0.00 
Wetness + Period 5 -193.91 398.25 1.86 
Wetness + ?15N + Period 6 -193.82 400.24 3.85 
Age + Sex + Wetness + Period 7 -195.30 405.40 9.00 
Age + Sex + Wetness + ?15N + Period 8 -195.36 407.75 11.35 
Age * Period + Sex * Period + Wetness 9 -195.81 410.92 14.53 
?15N + Period  5 -200.50 411.42 15.03 
Age * Period + Sex * Period + Wetness * 11 -193.79 411.53 15.14 
Period + ?15N 
Age * Period + Sex * Period + Wetness + 10 -195.83 413.27 16.88 
?15N 
Null 3 -203.97 414.10 17.71 
Age + Sex + ?15N + Period 7 -202.00 418.81 22.42 
     
Blood      
Wetness * Period 6 -215.19 442.91 0.00 
Wetness + ?15N + Period 6 -215.89 444.32 1.41 
Age * Period + Sex * Period + Wetness * 11 -210.39 444.51 1.60 
Period + ?15N 
Wetness + Period 5 -217.65 445.67 2.75 
Age * Period + Sex * Period + Wetness + 10 -213.09 447.62 4.70 
?15N 
Age * Period + Sex * Period + Wetness 9 -215.18 449.52 6.61 
Age + Sex + Wetness + ?15N + Period 8 -217.43 451.78 8.86 
?15N + Period 5 -220.72 451.82 8.91 
Age + Sex + Wetness + Period 7 -219.17 453.05 10.14 
Age + Sex + ?15N + Period 7 -222.24 459.19 16.27 
Null 3 -229.93 466.01 23.10 
 
 
136 
 
 
  
Table S5.2 AICc model selection for generalized mixed effect models describing 
variation in stable-carbon values of non-breeding wood thrush in Belize with known 
home ranges. Age (yearling or adult), sex (male or female), mean satellite-derived 
wetness of home ranges, ?15N (proxy for proportion of insects consumed) and 
sampling period were included as fixed effects. Separate models were run including 
isotope values derived from either claw (n = 97) which provides information on diets 
over a few months and blood (n = 110) which provides information on diets over a 
few weeks. All models included a random intercept for bird identity nested within 
year and models < 2 ?AICc were considered competitive. 
 
Model df Log- AICc ?AICc 
likelihood 
Claw     
Wetness * Period 6 -127.23 267.39 0.00 
Wetness + ?15N + Period 6 -128.30 269.53 2.14 
Wetness + Period 5 -129.65 269.97 2.58 
?15N + Period 5 -131.46 273.57 6.18 
Age + Sex + Wetness + ?15N + Period 8 -128.60 274.83 7.44 
Age + Sex + Wetness + Period 7 -130.17 275.60 8.21 
Age * Period + Sex * Period + Wetness * 11 -125.73 276.57 9.18 
Period + ?15N 
Age * Period + Sex * Period + Wetness + 10 -128.18 278.91 11.52 
?15N 
Age + Sex + ?15N + Period 7 -132.09 279.44 12.04 
Age * Period + Sex * Period + Wetness 9 -129.88 279.84 12.44 
Null  3 -137.01 280.27 12.88 
     
Blood      
Wetness * Period 6 -143.93 300.68 0.00 
Age * Period + Sex * Period + Wetness * 11 -138.14 300.97 0.29 
Period + ?15N 
Wetness + ?15N + Period 6 -144.87 302.55 1.88 
Wetness + Period 5 -146.84 304.26 3.58 
Age * Period + Sex * Period + Wetness + 10 -141.27 304.76 4.09 
?15N 
?15N + Period 5 -147.61 305.79 5.11 
Age * Period + Sex * Period + Wetness 9 -143.76 307.32 6.64 
Age + Sex + Wetness + ?15N + Period 8 -145.37 308.17 7.49 
Age + Sex + Wetness + Period 7 -147.53 310.16 9.48 
Age + Sex + ?15N + Period 7 -148.48 312.07 11.39 
Null 3 -155.19 316.61 15.93 
 
137 
 
 
  
Table S5.3 Model averaged parameter estimates, standard error and unconditional 
95% confidence intervals for top linear mixed effect models describing variation in 
stable-carbon values across the non-breeding season for wood thrush in Belize. 
Separate models were constructed for isotope values derived from claw which 
provides information on diets over a few months and blood which provides 
information on diets over a few weeks. Separate models were also constructed for 
wetness measured across plots (claws, n = 147; blood, n = 164) or home ranges (claw, 
n = 97; blood, n = 110). 
 
Parameter Estimate SE 95% CI 
Plot    
Claw    
Wetness -7.15 2.40 -11.85, -2.45 
Period -0.54 0.16 -0.85, -0.22 
Wetness * Period 0.29 3.12 -4.98, 5.4 
    
Blood    
Wetness -5.98 2.40 -10.69, -1.27 
Period -0.52 0.38 -1.26, 0.22 
Wetness * Period 3.02 3.06 -3.53, 8.04 
?15N 0.20 0.08 -0.13, 0.33 
Age (yearling) 0.00 0.16 -0.15, 0.15 
Sex (Male) 0.17 0.17 -0.17, 0.25 
Age (yearling) * Period -0.75 0.31 -0.86, 0.51 
Sex (Male) * Period -0.66 0.30 -0.77, 0.46 
    
Home range    
Claw    
Wetness -5.45 3.61 -13.00, 2.10 
Period -0.69 0.22 -1.15, -0.24 
Wetness * Period -0.01 4.54 -9.52, 9.50 
    
Blood    
Wetness -5.44 3.26 -11.82, 0.94 
Period -0.46 0.46 -1.36, 0.44 
Wetness * Period 4.78 4.32 -4.53, 12.43 
?15N 0.26 0.10 -0.14, 0.43 
Age (yearling) -0.10 0.21 -0.31, 0.24 
Sex (Male) 0.13 0.20 -0.22, 0.32 
Age (yearling) * Period -0.72 0.39 -1.12, 0.56 
Sex (Male) * Period -0.72 0.40 -1.12, 0.56 
Notes: Values in bold have confidence intervals that do not include zero. 
 
138 
 
 
  
Figure S5.1 Factors driving stable-carbon value in wood thrush across the non-
breeding season in Belize. Negative relationship between satellite-derived wetness on 
plots and stable-carbon values from (A) claws and (B) blood. More negative wetness 
values denote drier conditions. Line denotes the predicted relationship and shaded 
area the 95% CI. 
 
 
139 
 
 
  
Bibliography 
?kesson, S., and L.-A. Hansson, editors. 2014. Animal movement across scales. 
Oxford University Press, Oxford. 
?kesson, S., M. Ilieva, J. Karagicheva, E. Rakhimberdiev, B. Tomotani, and B. 
Helm. 2017. Timing avian long-distance migration: from internal clock 
mechanisms to global flights. Philosophical Transactions of the Royal Society B: 
Biological Sciences 372:20160252. 
Alerstam, T., A. Hedenstrom, and S. Akesson. 2003. Long-distance migration: 
evolution and determinants. Oikos 103:247?260. 
Alerstam, T., and ?. Lindstr?m. 1990. Optimal bird migration: the relative 
importance of time, energy and safety. Pages 331?351 in E. Gwinner, editor. 
Bird Migration: Physiology and Ecophysiology. Springer-Verlag, Berlin. 
Andr?n, H. 1994. Effects of Habitat Fragmentation on Birds and Mammals in 
Landscapes with Different Proportions of Suitable Habitat: A Review. Oikos 
71:355. 
Arnold, T. W. 2010. Uninformative Parameters and Model Selection Using Akaike?s 
Information Criterion. Journal of Wildlife Management 74:1175?1178. 
Bairlein, F. 2003. Nutritional Strategies in Migratory Birds BT  - Avian Migration. 
Pages 321?332 in P. Berthold, E. Gwinner, and E. Sonnenschein, editors. 
Springer Berlin Heidelberg, Berlin, Heidelberg. 
Balbont?n, J., A. P. M?ller, I. G. Hermosell, A. Marzal, M. Reviriego, and F. de Lope. 
2009. Individual responses in spring arrival date to ecological conditions during 
winter and migration in a migratory bird. The Journal of animal ecology 78:981?
140 
 
 
  
9. 
Barbaro, L., L. Couzi, V. Bretagnolle, J. Nezan, and F. Vetillard. 2008. Multi-scale 
habitat selection and foraging ecology of the eurasian hoopoe (Upupa epops) in 
pine plantations. Biodiversity and Conservation 17:1073?1087. 
Barton, K. 2017. MuMIn: Multi-Model Inference. https://cran.r-
project.org/package=MuMIn. 
Bates, D., M. M?chler, B. Bolker, and S. Walker. 2015. Fitting Linear Mixed-Effects 
Models Using lme4. Journal of Statistical Software 67. 
Bauchinger, U., and S. McWilliams. 2009. Carbon Turnover in Tissues of a Passerine 
Bird: Allometry, Isotopic Clocks, and Phenotypic Flexibility in Organ Size. 
Physiological and Biochemical Zoology 82:787?797. 
Bearhop, S., S. Waldron, S. C. Votier, and R. W. Furness. 2002. Factors That 
Influence Assimilation Rates and Fractionation of Nitrogen and Carbon Stable 
Isotopes in Avian Blood and Feathers 75:451?458. 
Beatty, W. S., E. B. Webb, D. C. Kesler, A. H. Raedeke, L. W. Naylor, and D. D. 
Humburg. 2014. Landscape effects on mallard habitat selection at multiple 
spatial scales during the non-breeding period. Landscape Ecology 29:989?1000. 
Berthold, P. 1975. Migratory fattening endogenous control and interaction with 
migratory activity. Naturwissenschaften:399. 
Berthold, P. 1978. Das Zusammenwirken von endogenen Zugzeit-Programmen und 
Umweltfaktoren beim Zugablauf bei Grasm?cken: Eine Hypothese. Die 
Vogelwarte 29:153?159. 
Berthold, P., E. Gwinner, and E. Sonnenschein, editors. 2003. Avian Migration. 
141 
 
 
  
Springer, Berlin. 
Besag, J., P. J. Diggle, S. Journal, R. Statistical, S. Series, and C. A. Statistics. 2016. 
Simple Monte Carlo Tests for Spatial Pattern Published by?: Wiley for the Royal 
Statistical Society Simple Monte Carlo Tests for Spatial Pattern 26:327?333. 
Bonnot, T. W., M. L. Wildhaber, J. J. Millspaugh, A. J. Delonay, R. B. Jacobson, and 
J. L. Bryan. 2011. Discrete choice modeling of shovelnose sturgeon habitat 
selection in the Lower Missouri River. Journal of Applied Ichthyology 27:291?
300. 
Boyle, W. A., D. R. Norris, and C. G. Guglielmo. 2010. Storms drive altitudinal 
migration in a tropical bird:2511?2519. 
Bridge, E. S., J. F. Kelly, P. E. Bjornen, C. M. Curry, P. H. C. Crawford, and J. M. 
Paritte. 2010. Effects of nutritional condition on spring migration: do migrants 
use resource availability to keep pace with a changing world? The Journal of 
experimental biology 213:2424?9. 
Brown, D. R., and T. W. Sherry. 2006a. Food supply controls the body condition of a 
migrant bird wintering in the tropics. Oecologia 149:22?32. 
Brown, D. R., and T. W. Sherry. 2008. Alternative strategies of space use and 
response to resource change in a wintering migrant songbird. Behavioral 
Ecology 19:1314?1325. 
Brown, D., and T. Sherry. 2006b. Behavioral response of resident jamaican birds to 
dry season food supplementation. Biotropica 38:91?99. 
Buler, J. J., and F. R. Moore. 2011. Migrant?habitat relationships during stopover 
along an ecological barrier: extrinsic constraints and conservation implications. 
142 
 
 
  
Journal of Ornithology 152:101?112. 
Buler, J. J., F. R. Moore, and S. Woltmann. 2007. A multi-scale examination of 
stopover habitat use by birds. Ecology 88:1789?1802. 
Burke, D., and E. Nol. 2000. Landscape and fragment size effects on reproductive 
success of forest-breeding birds in Ontario. Ecological Applications 10:1749?
1761. 
Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel 
inference: a practical information-theoretic approach. 2nd edition. Springer, New 
York. 
Calabrese, J. M., C. H. Fleming, and E. Gurarie. 2016. Ctmm: an R Package for 
Analyzing Animal Relocation Data As a Continuous-Time Stochastic Process. 
Methods in Ecology and Evolution 7:1124?1132. 
Calisi, R. M., and G. E. Bentley. 2009. Lab and field experiments: Are they the same 
animal? Hormones and Behavior 56:1?10. 
Catry, T., J. A. Alves, J. A. Gill, T. G. Gunnarsson, and J. P. Granadeiro. 2012. Sex 
Promotes Spatial and Dietary Segregation in a Migratory Shorebird during the 
Non-Breeding Season. PLoS ONE 7:e33811. 
Chandler, R. B., and D. I. King. 2011. Habitat quality and habitat selection of golden-
winged warblers in Costa Rica: an application of hierarchical models for open 
populations. Journal of Applied Ecology 48:1038?1047. 
Channan, S., K. Collins, and W. Emanuel. 2014. Global mosaics of the standard 
MODIS land cover type data. University of Maryland and the Pacific Northwest 
National Laboratory, College Park. 
143 
 
 
  
Cochran, W., and M. Wikelski. 2005. Individual migratory tactics of New World 
Catharus thrushes: Current knowledge and future tracking options from space. 
Pages 274?289 in R. Greenberg and P. P. Marra, editors. Birds of two worlds: 
the ecology and evolution of migration. Johns Hopkins University Press. 
Cody, M. 1985. An introduction to habitat selection in birds. Pages 3?56 in M. Cody, 
editor. Habitat Selection in Birds. Academic Press, Inc., Orlando. 
Cohen, E. B., J. A. Hostetler, M. T. Hallworth, C. S. Rushing, T. S. Sillett, and P. P. 
Marra. 2018. Quantifying the strength of migratory connectivity. Methods in 
Ecology and Evolution 9:513?524. 
Cohen, W., and S. Goward. 2014. Landsat?s Role in Ecological Applications of 
Remote Sensing 54:535?545. 
Conklin, J. R., P. F. Battley, and M. a Potter. 2013. Absolute consistency: individual 
versus population variation in annual-cycle schedules of a long-distance migrant 
bird. PloS one 8:e54535. 
Cooper, A. B., and J. J. Millspaugh. 1999. the Application of Discrete Choice Models 
To Wildlife. Ecology 80:566?575. 
Cooper, N. W., T. W. Sherry, and P. P. Marra. 2015. Experimental reduction of 
winter food decreases body condition and delays migration in a long-distance 
migratory bird. Ecology 96:1933?1942. 
Cornelius, J. M., T. Boswell, S. Jenni-Eiermann, C. W. Breuner, and M. Ramenofsky. 
2013. Contributions of endocrinology to the migration life history of birds. 
General and Comparative Endocrinology 190:47?60. 
Cox, D. R. 1972. Regression Models and Life-Tables. Journal of the Royal Statistical 
144 
 
 
  
Society. Series B (Methodological) 34:187?220. 
Crist, E. P., and R. J. Kauth. 1986. The Tasseled Cap De-Mystified:81?86. 
Diggs, N. E., P. P. Marra, and R. J. Cooper. 2011. Resource Limitation Drives 
Patterns of Habitat Occupancy during The Nonbreeding Season for an 
Omnivorous Songbird. The Condor 113:646?654. 
Dodge, S., G. Bohrer, R. Weinzierl, S. C. Davidson, R. Kays, D. Douglas, S. Cruz, J. 
Han, D. Brandes, and M. Wikelski. 2013. The environmental-data automated 
track annotation (Env-DATA) system: linking animal tracks with environmental 
data. Movement Ecology 1:3. 
Donovan, T., R. Lamberson, A. Kimber, F. I. Thompson, and J. Faaborg. 1995. 
Modeling the effects of habitat fragmentation on source and sink demography of 
Neotropical migrant birds. Conservation Biology 9:1396?1407. 
Doren, B. M. Van, M. Liedvogel, and B. Helm. 2017. Programmed and flexible: 
long-term Zugunruhe data highlight the many axes of variation in avian 
migratory behaviour. Journal of Avian Biology 48:155?172. 
Drake, A., C. Rock, S. P. Quinlan, and D. J. Green. 2013. Carry-over effects of winter 
habitat vary with age and sex in yellow warblers Setophaga petechia. Journal of 
Avian Biology 44:321?330. 
Dray, S., and A.-B. Dufour. 2007. The ade4 Package: Implementing the Duality 
Diagram for Ecologists. Journal of Statistical Software 22. 
Eikenaar, C., F. Bairlein, M. St?we, and S. Jenni-Eiermann. 2014. Corticosterone, 
food intake and refueling in a long-distance migrant. Hormones and behavior 
65:480?7. 
145 
 
 
  
Ekstr?m, C. 2019. MESS: Miscellaneous Esoteric Statistical Scripts. 
https://CRAN.R-project.org/package=MESS. 
Evans, M., E. Gow, R. Roth, M. Johnson, and T. Underwood. 2011. Wood Thrush 
(Hylocichla mustelina). Page The Birds of North America. Version 2. Cornell 
Lab of Ornithology, Ithaca, NY. 
Faaborg, J., R. Holmes, A. D. Anders, K. L. Bildstein, K. M. Dugger, S. A. 
Gauthreaux, P. Heglund, K. A. Hobson, A. E. Jahn, D. H. Johnson, S. C. Latta, 
D. J. Levey, P. P. Marra, C. Merkord, E. Nol, S. I. Rothstein, T. Sherry, T. S. 
Sillett, F. I. Thompson, and N. Warnock. 2010. Conserving migratory land birds 
in the New World: Do we know enough? Ecological Applications 20:398?418. 
Fauth, P. 2001. Wood Thrush populations are not all sinks in the agricultural 
midwestern United States. Conservation Biology 15:523?527. 
Finch, T., S. J. Butler, A. M. A. Franco, and W. Cresswell. 2017. Low migratory 
connectivity is common in long-distance migrant birds. Journal of Animal 
Ecology 86:662?673. 
Fleming, C. H., and J. M. Calabrese. 2017. A new kernel density estimator for 
accurate home-range and species-range area estimation. Methods in Ecology and 
Evolution 8:571?579. 
Fraser, K. C., A. Shave, A. Savage, A. Ritchie, K. Bell, J. Siegrist, J. D. Ray, K. 
Applegate, and M. Pearman. 2017. Determining fine-scale migratory 
connectivity and habitat selection for a migratory songbird by using new GPS 
technology. Journal of Avian Biology 48:339?345. 
Fretwell, S. D., and H. L. Lucas. 1969. On territorial behavior and other factors 
146 
 
 
  
influencing habitat distribution in birds. Acta Biotheoretica 19:16?36. 
Friedl, M. A., D. Sulla-Menashe, B. Tan, A. Schneider, N. Ramankutty, A. Sibley, 
and X. Huang. 2010. MODIS Collection 5 global land cover: Algorithm 
refinements and characterization of new datasets. Remote Sensing of 
Environment 114:168?182. 
Fusani, L., M. Cardinale, I. Schwabl, and W. Goymann. 2011. Food availability but 
not melatonin affects nocturnal restlessness in a wild migrating passerine. 
Hormones and behavior 59:187?92. 
Gauthreaux, S. A. 1996. Bird Migration: Methodologies and Major Research 
Trajectories (1945-1995). The Condor 98:442?453. 
Gelman, A., J. Carlin, H. Stern, and D. Rubin. 2004. Bayesian Data Analysis. Second 
Edi. CRC/Chapman and Hall, Boca Raton. 
Gelman, A., X. Meng, and H. Stern. 1996. Posterior predictive assessment of model 
fitness via realized discrepancies. Statistica Sinica 6:733?760. 
Gilroy, J. J., J. A. Gill, S. H. M. Butchart, V. R. Jones, and A. M. A. Franco. 2016. 
Migratory diversity predicts population declines in birds. Ecology Letters 
19:308?317. 
G?mez, C., N. J. Bayly, D. R. Norris, S. A. Mackenzie, K. V. Rosenberg, P. D. 
Taylor, K. A. Hobson, and C. Daniel Cadena. 2017. Fuel loads acquired at a 
stopover site influence the pace of intercontinental migration in a boreal 
songbird. Scientific Reports 7:1?11. 
Gosler, A. 1991. On the use of greater covert moult and pectoral muscle as measures 
of condition in passerines with data for the Great Tit Parsus major. Bird Study 
147 
 
 
  
38:1?9. 
Gosler, A. 1996. Environmental and social determinants of winter fat storage in the 
great tit Parus major. Journal of Animal Ecology 65:1?17. 
Graham, C. H., and J. G. Blake. 2001. Influence of patch- and landscape-level factors 
on bird assemblages in a fragmented tropical landscape. Ecological Applications 
11:1709?1721. 
Grand, J., and S. A. Cushman. 2003. A multi-scale analysis of species-environment 
relationships: breeding birds in a pitch pine?scrub oak (Pinus rigida?Quercus 
ilicifolia) community. Biological Conservation 112:307?317. 
Greenberg, R., and V. Salewski. 2005. Ecological correlates of wintering social 
systems in New World and Old World migratory passerines. Pages 333?358 
Birds of Two Worlds: The Ecology and Evolution of Migration. Johns Hopkins 
University Press, Baltimore. 
Griffiths, R., M. C. Double, K. Orr, and R. J. G. Dawson. 1998. A DNA test to sex 
most birds. Molecular Ecology 7:1071?1075. 
Gwinner, E. 1996. Circadian and circannual programmes in avian migration. The 
Journal of experimental biology 199:39?48. 
Gwinner, E., J. P. Dittami, and H. J. A. Beldhuis. 1988. The seasonal development of 
photoperiodic responsiveness in an equatorial migrant, the garden warbler Sylvia 
borin. Journal of Comparative Physiology A 162:389?396. 
Harrison, X. a, J. D. Blount, R. Inger, D. R. Norris, and S. Bearhop. 2011. Carry-over 
effects as drivers of fitness differences in animals. The Journal of animal 
ecology 80:4?18. 
148 
 
 
  
Heckscher, C. M., M. R. Halley, and P. M. Stampul. 2015. Intratropical migration of 
a Nearctic-Neotropical migratory songbird (Catharus fuscescens) in South 
America with implications for migration theory. Journal of Tropical Ecology 
31:285?289. 
Helm, B., M. E. Visser, W. Schwartz, N. Kronfeld-Schor, M. Gerkema, T. Piersma, 
and G. Bloch. 2017. Two sides of a coin: ecological and chronobiological 
perspectives of timing in the wild. Philosophical Transactions of the Royal 
Society B: Biological Sciences 372:20160246. 
Hijmans, R. 2016. raster: Geographic Data Analysis and Modeling. https://cran.r-
project.org/package=raster. 
Hijmans, R. 2017. geosphere: Spherical Trigonometry. https://CRAN.R-
project.org/package=geosphere. 
Hobson, K. a. 2011. Isotopic ornithology: a perspective. Journal of Ornithology 
152:49?66. 
Hobson, K. A. 1993. Trophic relationships among high Arctic seabirds: insights from 
tissue-dependent stable-isotope models. Marine Ecology Progress Series 95:7?
18. 
Holmes, R., T. Sherry, and L. Reitsma. 1989. Population structure, territoriality and 
overwinter survival of two migrant warbler species in Jamaica. Condor 91:545?
561. 
Hutto, R. 1985. Habitat selection by nonbreeding, migratory land birds. Pages 455?
476 in M. Cody, editor. Habitat Selection in Birds. Academic Press, Inc., 
Orlando. 
149 
 
 
  
Janzen, D. 1973. Sweep samples of tropical foliage insects: effects of seasons, 
vegetation types, elevation, time of day, and insularity. Ecology 54:687?708. 
Janzen, D. H., and T. W. Schoener. 1968. Differences in Insect Abundance and 
Diversity Between Wetter and Drier Sites During a Tropical Dry Season. 
Ecology 49:96?110. 
Johnson, D. H. 1980. The Comparison of Usage and Availability Measurements for 
Evaluating Resource Preference. Ecology 61:65?71. 
Jones, J. 2001. Habitat Selection Studies in Avian Ecology: A Critical Review. The 
Auk 118:557. 
Kaiser, S. A., and C. A. Lindell. 2007. Effects of distance to edge and edge type on 
nestling growth and nest survival in the wood thrush. The Condor 109:288?303. 
Kellner, K. 2017. jagsUI: A Wrapper Around ?rjags? to Streamline ?JAGS? 
Analyses. https://cran.r-project.org/package=jagsUI. 
Kelly, J. F., F. R. Moore, R. J. Smith, D. M. Finch, and W. Yong. 1999. Influence of 
Summer Biogeography on Wood Stopover Abundance. The Condor:76?85. 
Ketterson, E. D., and V. Nolan. 1983. The Evolution of Differential Bird Migration. 
Pages 357?402 Current Ornithology. Springer US, Boston, MA. 
Killick, R., and I. A. Eckley. 2015. changepoint?: An R Package for Changepoint 
Analysis . Journal of Statistical Software 58:1?15. 
Knight, S. M., D. W. Bradley, R. G. Clark, E. A. Gow, M. B?lisle, L. L. Berzins, T. 
Blake, E. S. Bridge, L. Burke, R. D. Dawson, P. O. Dunn, D. Garant, G. L. 
Holroyd, D. J. T. Hussell, O. Lansdorp, A. J. Laughlin, M. L. Leonard, F. 
Pelletier, D. Shutler, L. Siefferman, C. M. Taylor, H. E. Trefry, C. M. Vleck, D. 
150 
 
 
  
Vleck, D. W. Winkler, L. A. Whittingham, and D. R. Norris. 2018. Constructing 
and evaluating a continent-wide migratory songbird network across the annual 
cycle. Ecological Monographs 88:445?460. 
Kohn, M. J. 2010. Carbon isotope compositions of terrestrial C3 plants as indicators 
of (paleo)ecology and (paleo)climate. Proceedings of the National Academy of 
Sciences 107:19691?19695. 
Kokko, H. 1999. Competition for early arrival in migratory birds. Journal of Animal 
Ecology 68:940?950. 
Kole?ek, J., S. Hahn, T. Emmenegger, and P. Proch?zka. 2018. Intra-tropical 
movements as a beneficial strategy for Palearctic migratory birds. Royal Society 
Open Science 5:171675. 
Kole?ek, J., P. Proch?zka, N. El-Arabany, M. Tarka, M. Ilieva, S. Hahn, M. Honza, J. 
de la Puente, A. Bermejo, A. G?rsoy, S. Bensch, P. Zehtindjiev, D. Hasselquist, 
and B. Hansson. 2016. Cross-continental migratory connectivity and 
spatiotemporal migratory patterns in the great reed warbler. Journal of Avian 
Biology 47:756?767. 
Kranstauber, B., M. Smolla, and A. Scharf. 2018. move: Visualizing and Analyzing 
Animal Track Data. https://CRAN.R-project.org/package=move. 
Kutner, M., C. Nachtscheim, and J. Neter. 2004. Applied Linear Regression Models-
Revised Edition. 4th edition. McGraw-Hill, Boston. 
Lafleur, J. M., J. J. Buler, and F. R. Moore. 2016. Geographic position and landscape 
composition explain regional patterns of migrating landbird distributions during 
spring stopover along the northern coast of the Gulf of Mexico. Landscape 
151 
 
 
  
Ecology 31:1697?1709. 
Lang, J. D., L. A. Powell, D. G. Krementz, and M. J. Conroy. 2002. Wood thrush 
movements and habitat use: Effects of forest management for red-cockaded 
woodpeckers. Auk 119:109?124. 
Latta, S., and J. Faaborg. 2002. Demographic and population responses of Cape May 
Warblers wintering in multiple habitats. Ecology 83:2502?2515. 
Lee, M., L. Fahrig, K. Freemark, and D. J. Currie. 2002. Importance of patch scale vs 
landscape scale on selected forest birds. Oikos 96:110?118. 
Lenth, R. 2018. emmeans: Estimated Marginal Means, aka Least-Squares Means. 
https://cran.r-project.org/package=emmeans. 
Lessells, C. M., and P. T. Boag. 1987. Unrepeatable Repeatabilities: A Common 
Mistake. The Auk 104:116?121. 
Li, P., L. Jiang, and Z. Feng. 2013. Cross-comparison of vegetation indices derived 
from landsat-7 enhanced thematic mapper plus (ETM+) and landsat-8 
operational land imager (OLI) sensors. Remote Sensing 6:310?329. 
Lindell, C. A., S. K. Riffell, S. A. Kaiser, A. L. Battin, M. L. Smith, and T. D. Sisk. 
2007. Edge Responses of Tropical and Temperate Birds. The Wilson Journal of 
Ornithology 119:205?220. 
Lindstr?m, ?., B. B. Chapman, N. Jonz?n, and M. Klaassen. 2014. Movement and 
migration in a changing world. Pages 36?50 Animal Movement Across Scales. 
Oxford University Press. 
Lofts, B., A. J. Marshall, and A. Wolfson. 2008. The experimental demonstation of 
pre-migration activity in the absence of fat deposition in birds. Ibis 105:99?105. 
152 
 
 
  
Louren?o, P. M., J. P. Granadeiro, J. L. Guilherme, and T. Catry. 2015. Turnover 
rates of stable isotopes in avian blood and toenails: Implications for dietary and 
migration studies. Journal of Experimental Marine Biology and Ecology 472:89?
96. 
Maggini, I., and F. Bairlein. 2010. Endogenous rhythms of seasonal migratory body 
mass changes and nocturnal restlessness in different populations of Northern 
Wheatear Oenanthe oenanthe. Journal of biological rhythms 25:268?76. 
Manly, B. F., L. McDonald, D. Thomas, T. L. McDonald, and W. P. Erickson. 2004. 
Resource Selection by Animals. 2nd edition. Kluwer Academic Publishers, 
Dordrecht. 
Marra, P. P. 2000. The role of behavioral dominance in structuring patterns of habitat 
occupancy in a migrant bird during the nonbreeding season. Behavioral Ecology 
11:299?308. 
Marra, P. P., E. B. Cohen, S. R. Loss, J. E. Rutter, and C. M. Tonra. 2015. A call for 
full annual cycle research in animal ecology. Biology Letters 11:20150552. 
Marra, P. P., K. A. Hobson, and R. T. Holmes. 1998. Linking Winter and Summer 
Events in a Migratory Bird by Using Stable-Carbon Isotopes. Science 282:1884?
1886. 
Mayor, S. J., D. C. Schneider, J. A. Schaefer, and S. P. Mahoney. 2009. Habitat 
selection at multiple scales. ?coscience 16:238?247. 
Mccabe, J. D., and B. J. Olsen. 2015. Landscape-scale habitat availability, and not 
local geography, predicts migratory landbird stopover across the Gulf of Maine. 
Journal of Avian Biology 46:395?405. 
153 
 
 
  
McFadden, D. 1978. Modeling the choice of residential location. Pages 410?428 in 
A. Karlqvist, L. Lundqvist, F. Snickars, and J. Weibull, editors. Spatial 
interaction theory and planning models. North Holland Publishing Company, 
Amsterdam. 
McGarigal, K., H. Y. Wan, K. A. Zeller, B. C. Timm, and S. A. Cushman. 2016. 
Multi-scale habitat selection modeling: a review and outlook. Landscape 
Ecology 31:1161?1175. 
McKellar, A. E., P. P. Marra, S. J. Hannon, C. E. Studds, and L. M. Ratcliffe. 2013a. 
Winter rainfall predicts phenology in widely separated populations of a migrant 
songbird. Oecologia 172:595?605. 
McKellar, A. E., P. P. Marra, and L. M. Ratcliffe. 2013b. Starting over: experimental 
effects of breeding delay on reproductive success in early-arriving male 
American redstarts. Journal of Avian Biology 44:495?503. 
McKinnon, E. A., J. A. Rotenberg, and B. J. M. Stutchbury. 2015a. Seasonal change 
in tropical habitat quality and body condition for a declining migratory songbird. 
Oecologia 179:363?375. 
McKinnon, E. A., C. Q. Stanley, and B. J. M. Stutchbury. 2015b. Carry-Over Effects 
of Nonbreeding Habitat on Start-to-Finish Spring Migration Performance of a 
Songbird. PLOS ONE 10:e0141580. 
McKinnon, E., K. Fraser, and B. Stutchbury. 2013. New discoveries in landbird 
migration using geolocators, and a flight plan for the future. The Auk 130:211?
222. 
McNamara, J. M., and A. I. Houston. 1990. The value of fat reserves and the tradeoff 
154 
 
 
  
between starvation and predation. Acta Biotheoretica 38:37?61. 
Meyer, C. B., and W. Thuiller. 2006. Accuracy of resource selection functions across 
spatial scales. Diversity & Distributions 12:288?297. 
Michener, R., and K. Lajtha, editors. 2007. Stable Isotopes in Ecology and 
Environmental Science. Blackwell Publishing Ltd, Oxford, UK. 
Milner-Gulland, E. J., J. M. Fryxell, and A. R. E. Sinclair, editors. 2011. Animal 
Migration. Oxford University Press. 
Mitchell, M. S., R. A. Lancia, and J. A. Gerwin. 2001. Using Landscape-Level Data 
to Predict the Distribution of Birds on a Managed Forest: Effects of Scale. 
Ecological Applications 11:1692?1708. 
M?ller, A. 1994. Phenotype-dependent arrival time and its consequences in a 
migratory bird. Behavioral Ecology and Sociobiology 35:115?122. 
Moore, F., and D. Aborn. 2000. Mechanisms of En Route Habitat Selection: How Do 
Migrants Make Habitat Decisions During Stopover? Studies in Avian Biology 
20:34?42. 
Moore, F. R., S. a. Gauthreaux, Jr., P. Kerlinger, and T. R. Simons. 1993. Stopover 
habitat: management implications and guidelines. Status and Management of 
Neotropical Migratory Birds:58?69. 
Moreau, R. 1972. The Palaearctic-African Bird Migration Systems. Acadmic Press 
Inc, London. 
Morton, E. 1980. Adaptations to seasonal changes by migrant land birds in the 
Panama Canal Zone. Pages 437?453 in A. Keast and E. Morton, editors. Migrant 
Birds in the Neotropics: Ecology, Behavior, Distribution, and Conservation. 
155 
 
 
  
Smithsonian Institution Press, Washington, DC. 
Muheim, R., I. Henshaw, S. Sj?berg, and M. E. Deutschlander. 2014. BirdOriTrack: 
A new video-tracking program for orientation research with migratory birds. 
Journal of Field Ornithology 85:91?105. 
Neelin, J. D., M. M?nnich, H. Su, J. E. Meyerson, and C. E. Holloway. 2006. 
Tropical drying trends in global warming models and observations. Proceedings 
of the National Academy of Sciences of the United States of America 103:6110?
5. 
Newton, I. 2008. The migration ecology of birds. Academic Press/Elsevier, 
Amsterdam. 
Nilsson, C., R. H. G. Klaassen, and T. Alerstam. 2013. Differences in Speed and 
Duration of Bird Migration between Spring and Autumn. The American 
Naturalist 181:837?845. 
Norevik, G., G. Boano, A. Hedenstr?m, R. Lardelli, F. Liechti, and S. ?kesson. 2018. 
Highly mobile insectivorous swifts perform multiple intra-tropical migrations to 
exploit an asynchronous African phenology. Oikos:1?9. 
Norris, D. R., P. P. Marra, T. K. Kyser, T. W. Sherry, and L. M. Ratcliffe. 2004. 
Tropical winter habitat limits reproductive success on the temperate breeding 
grounds in a migratory bird. Proceedings of the Royal Society B: Biological 
Sciences 271:59?64. 
Oksanen, J., F. G. Blanchet, M. Friendly, R. Kindt, P. Legendre, D. McGlinn, P. 
Minchin, R. O?Hara, G. Simpson, P. Solymos, M. Stevens, E. Szoecs, and H. 
Wagner. 2017. vegan: Community Ecology Package. https://cran.r-
156 
 
 
  
project.org/package=vegan. 
Orians, G. H., and J. F. Wittenberger. 1991. Spatial and Temporal Scales in Habitat 
Selection Published by?: The University of Chicago P. The American Naturalist 
137:S29?S49. 
Pearson, S. F., D. J. Levey, C. H. Greenberg, and C. Mart?nez Del Rio. 2003. Effects 
of elemental composition on the incorporation of dietary nitrogen and carbon 
isotopic signatures in an omnivorous songbird. Oecologia 135:516?23. 
Petit, D. R. 2000. Habitat Use by Landbirds along Nearctic-Neotropical Migration 
Routes: Implications for Conservation of Stopover Traits. Studies in Avian 
Biology 20:15?30. 
Pickens, B. A., S. L. King, P. L. Vasseur, S. E. Zimorski, and W. Selman. 2017. 
Seasonal Movements and Multiscale Habitat Selection of Whooping Crane ( 
Grus americana ) in Natural and Agricultural Wetlands. Waterbirds 40:322?333. 
Pinheiro, J., D. Bates, S. DebRoy, D. Sarkar, and R. C. Team. 2017. nlme: Linear and 
Nonlinear Mixed Effects Models. https://CRAN.R-project.org/package=nlme. 
Plummer, M. 2003. JAGS: A program for analysis of Bayesian graphical models 
using Gibbs sampling. Proceedings of the 3rd International Workshop on 
Distributed Statistical Computing:1?10. 
R Core Team, R. C. 2017. R: A Language and Environment for Statistical 
Computing. Vienna, Austria. 
Ramenofsky, M., J. M. Cornelius, and B. Helm. 2012. Physiological and behavioral 
responses of migrants to environmental cues. Journal of Ornithology 153:181?
191. 
157 
 
 
  
Ramenofsky, M., and J. C. Wingfield. 2017. Regulation of complex behavioural 
transitions: migration to breeding. Animal Behaviour 124:299?306. 
Rappole, J., M. Ramos, and K. Winker. 1989. Wintering Wood Thrush movements 
and mortality in southern Veracruz. The Auk 106:402?410. 
Rappole, J., and A. Tipton. 1991. New Harness Design for Attachment of Radio 
Transmitters to Small Passerines. Journal of field ornithology 62:335?337. 
Renfrew, R. B., D. Kim, N. Perlut, J. Smith, J. Fox, and P. P. Marra. 2013. 
Phenological matching across hemispheres in a long-distance migratory bird. 
Diversity and Distributions 19:1008?1019. 
Rettie, W. J., and F. Messier. 2000. Hierarchical habitat selection by woodland 
caribou: its relationship to limiting factors. Ecography 23:466?478. 
Reudink, M. W., P. P. Marra, T. K. Kyser, P. T. Boag, K. M. Langin, and L. M. 
Ratcliffe. 2009. Non-breeding season events influence sexual selection in a long-
distance migratory bird. Proceedings of the Royal Society B: Biological 
Sciences 276:1619?26. 
Riitters, K. H., J. D. Wickham, R. V. O?Neill, K. B. Jones, E. R. Smith, J. W. 
Coulston, T. G. Wade, and J. H. Smith. 2002. Fragmentation of continental 
United States forests. Ecosystems 5:815?822. 
Roberts, D. L. 2011. Conservation value of forest fragments for wood thrsuh ( 
Hylocichla mustelina ) in Costa Rica?s caribbean lowlands. Latin American 
Journal of Conservation 2:8?17. 
Robinson, S. K., F. R. Thompson, T. M. Donovan, D. R. Whitehead, and J. Faaborg. 
1995. Regional Forest Fragmentation and the Nesting Success of Migratory 
158 
 
 
  
Birds. Science 267:1987?1990. 
Rockwell, S. M., J. M. Wunderle, T. S. Sillett, C. I. Bocetti, D. N. Ewert, D. Currie, J. 
D. White, and P. P. Marra. 2017. Seasonal survival estimation for a long-
distance migratory bird and the influence of winter precipitation. Oecologia 
183:715?726. 
Rushing, C. S., M. R. Dudash, C. E. Studds, and P. P. Marra. 2015. Annual variation 
in long-distance dispersal driven by breeding and non-breeding season climatic 
conditions in a migratory bird. Ecography 38:1006?1014. 
Rushing, C. S., T. B. Ryder, A. L. Scarpignato, J. F. Saracco, and P. P. Marra. 2016. 
Using demographic attributes from long-term monitoring data to delineate 
natural population structure. Journal of Applied Ecology 53:491?500. 
Saino, N., R. Ambrosini, M. Caprioli, F. Liechti, A. Romano, D. Rubolini, and C. 
Scandolara. 2017. Wing morphology, winter ecology, and fecundity selection: 
evidence for sex-dependence in barn swallows (Hirundo rustica). Oecologia 
184:799?812. 
Saino, N., T. Sz?p, R. Ambrosini, M. Romano, and A. P. M?ller. 2004. Ecological 
conditions during winter affect sexual selection and breeding in a migratory bird. 
Proceedings of the Royal Society B: Biological Sciences 271:681?6. 
Sauer, J. R., W. A. Link, J. E. Fallon, K. L. Pardieck, and D. J. Ziolkowski. 2013. The 
North American Breeding Bird Survey 1966?2011: Summary Analysis and 
Species Accounts. North American Fauna 79:1?32. 
Schlaich, A. E., R. H. G. Klaassen, W. Bouten, V. Bretagnolle, B. J. Koks, A. Villers, 
and C. Both. 2016. How individual Montagu?s Harriers cope with Moreau?s 
159 
 
 
  
Paradox during the Sahelian winter. Journal of Animal Ecology 85:1491?1501. 
Sexton, J. O., X. P. Song, M. Feng, P. Noojipady, A. Anand, C. Huang, D. H. Kim, 
K. M. Collins, S. Channan, C. DiMiceli, and J. R. Townshend. 2013. Global, 30-
m resolution continuous fields of tree cover: Landsat-based rescaling of MODIS 
vegetation continuous fields with lidar-based estimates of error. International 
Journal of Digital Earth 6:427?448. 
Shannon, C. E. 1948. A mathematical theory of communication. The Bell System 
Technical Journal 27:379?423. 
She, X., L. Zhang, Y. Cen, T. Wu, C. Huang, and M. H. A. Baig. 2015. Comparison 
of the continuity of vegetation indices derived from Landsat 8 OLI and Landsat 
7 ETM+ data among different vegetation types. Remote Sensing 7:13485?
13506. 
Sherry, T., M. Johnson, and A. Strong. 2005. Does winter food limit populations of 
migratory birds? Pages 414?425 in R. Greenberg and P. Marra, editors. Birds of 
Two Worlds: The Ecology and Evolution of Migration. Johns Hopkins 
University Press, Baltimore, MD. 
Smith, H. G., and J.-?. Nilsson. 1987. Intraspecific Variation in Migratory Pattern of 
a Partial Migrant, the Blue Tit (Parus caeruleus): An Evaluation of Different 
Hypotheses. The Auk 104:109?115. 
Smith, J. A. M., L. R. Reitsma, and P. P. Marra. 2011a. Influence of Moisture and 
Food Supply on the Movement Dynamics of a Nonbreeding Migratory Bird ( 
Parkesia noveboracensis ) in a Seasonal Landscape. The Auk 128:43?52. 
Smith, J., L. Reitsma, and P. Marra. 2011b. Influence of moisture and food supply on 
160 
 
 
  
the movement dynamics of a nonbreeding migratory bird (Parkesia 
noveboracensis) in a seasonal landscape. The Auk 128:43?52. 
La Sorte, F. A., M. W. Tingley, and A. H. Hurlbert. 2014. The role of urban and 
agricultural areas during avian migration: An assessment of within-year 
temporal turnover. Global Ecology and Biogeography 23:1225?1234. 
Stanley, C. Q., M. MacPherson, K. C. Fraser, E. A. McKinnon, and B. J. M. 
Stutchbury. 2012. Repeat Tracking of Individual Songbirds Reveals Consistent 
Migration Timing but Flexibility in Route. PLoS ONE 7:e40688. 
Stanley, C. Q., E. A. Mckinnon, K. C. Fraser, M. P. Macpherson, G. Casbourn, L. 
Friesen, P. P. Marra, C. Studds, T. B. Ryder, N. E. Diggs, and B. J. M. 
Stutchbury. 2015. Connectivity of wood thrush breeding, wintering, and 
migration sites based on range-wide tracking. Conservation Biology 29:164?
174. 
Stoffel, M. A., S. Nakagawa, and H. Schielzeth. 2017. rptR: repeatability estimation 
and variance decomposition by generalized linear mixed-effects models. 
Methods in Ecology and Evolution 8:1639?1644. 
Strong, A. M., and T. W. Sherry. 2000. Habitat-specific effects of food abundance on 
the condition of ovenbirds wintering in Jamaica. Journal of Animal Ecology 
69:883?895. 
Studds, C. E., T. K. Kyser, and P. P. Marra. 2008. Natal dispersal driven by 
environmental conditions interacting across the annual cycle of a migratory 
songbird. Proceedings of the National Academy of Sciences of the United States 
of America 105:2929?33. 
161 
 
 
  
Studds, C. E., and P. P. Marra. 2005. Nonbreeding habitat occupancy and population 
processes: An upgrade experiment with a migratory bird. Ecology 86:2380?
2385. 
Studds, C. E., and P. P. Marra. 2011a. Reconsidering the role of precipitation and 
food availability in relation to the effect of photoperiod on spring departure of a 
migratory bird. Proceedings of the Royal Society B: Biological Sciences 
279:17?18. 
Studds, C. E., and P. P. Marra. 2011b. Rainfall-induced changes in food availability 
modify the spring departure programme of a migratory bird. Proceedings of the 
Royal Society B: Biological Sciences 278:3437?3443. 
Studds, C., and P. Marra. 2007. Linking fluctuations in rainfall to nonbreeding season 
performance in a long-distance migratory bird, Setophaga ruticilla. Climate 
Research 35:115?122. 
Stutchbury, B. J. M., R. Siddiqui, K. Applegate, G. T. Hvenegaard, P. Mammenga, N. 
Mickle, M. Pearman, J. D. Ray, A. Savage, T. Shaheen, and K. C. Fraser. 2016. 
Ecological Causes and Consequences of Intratropical Migration in Temperate-
Breeding Migratory Birds. The American Naturalist 188:S28?S40. 
Symonds, M. R. E., and A. Moussalli. 2011. A brief guide to model selection, 
multimodel inference and model averaging in behavioural ecology using 
Akaike?s information criterion. Behavioral Ecology and Sociobiology 65:13?21. 
Terrill, S. B., and R. D. Ohmart. 1984. Facultative Extension of Fall Migration by 
Yellow-Rumped Warblers (Dendroica coronata). The Auk 101:427?438. 
Therneau, T. M. 2015. A Package for Survival Analysis in S. https://CRAN.R-
162 
 
 
  
project.org/package=survival. 
Therneau, T. M., and P. M. Grambsch. 2000. Modeling Survival Data: Extending the 
Cox Model. Springer New York, New York, NY. 
Thomas, D. L., D. Johnson, and B. Griffith. 2006. A Bayesian Random Effects 
Discrete-Choice Model for Resource Selection?: Population-Level Selection 
Inference. Journal of Wildlife Management 70:404?412. 
Thompson III, F. R., T. M. Donovan, R. M. DeGraff, J. Faaborg, and S. K. Robinson. 
2002. A multi-scale perspective of the effects of forest fragmentation on birds in 
eastern forests. Studies in Avian Biology 25:8?19. 
Thorup, K., A. P. T?ttrup, M. Willemoes, R. H. G. Klaassen, R. Strandberg, M. L. 
Vega, H. P. Dasari, M. B. Ara?jo, M. Wikelski, and C. Rahbek. 2017. Resource 
tracking within and across continents in long-distance bird migrants. Science 
Advances 3:e1601360. 
Totzke, U., A. H?binger, J. Dittami, and F. Bairlein. 2000. The autumnal fattening of 
the long-distance migratory garden warbler (Sylvia borin) is stimulated by 
intermittent fasting. Journal of Comparative Physiology - B Biochemical, 
Systemic, and Environmental Physiology 170:627?631. 
Trierweiler, C., W. C. Mulli?, R. H. Drent, K. M. Exo, J. Komdeur, F. Bairlein, A. 
Harouna, M. De Bakker, and B. J. Koks. 2013. A Palaearctic migratory raptor 
species tracks shifting prey availability within its wintering range in the Sahel. 
Journal of Animal Ecology 82:107?120. 
Valente, J. J., and M. G. Betts. 2018. Response to fragmentation by avian 
communities is mediated by species traits. Diversity and Distributions:1?13. 
163 
 
 
  
Vega Rivera, J., J. H. Rappole, W. J. McShea, and C. A. Haas. 1998. Wood Thrush 
Postfledging Movements and Habitat Use in Northern Virginia. The Condor 
100:69?78. 
Vernasco, B. J., T. S. Sillett, P. P. Marra, and T. B. Ryder. 2018. Environmental 
predictors of nestling condition, postfledging movement, and postfledging 
survival in a migratory songbird, the Wood Thrush ( Hylocichla mustelina ). The 
Auk 135:15?24. 
Villard, M. A., and J. P. Metzger. 2014. Beyond the fragmentation debate: A 
conceptual model to predict when habitat configuration really matters. Journal of 
Applied Ecology 51:309?318. 
Webster, M. S., and P. P. Marra. 2005. The importance of understanding migratory 
connectivity and seasonal interactions. Pages 199?209 in R. Greenberg and P. P. 
Marra, editors. Birds of Two Worlds: The Ecology and Evolution of Migration. 
Johns Hopkins University Press, Baltimore. 
Weinberg, H., and R. Roth. 1998. Forest area and habitat quality for nesting Wood 
Thrushes. The Auk 115:879?889. 
Wiens, J. A. 1989. Spatial Scaling in Ecology. Functional Ecology 3:385. 
Wiens, J. a, and B. T. Milne. 1989. Scaling of ? landscapes ? in landscape ecology , or 
, landscape ecology from a beetle ? s perspective. Landscape Ecology 3:87?96. 
Wilcove, D. S., and M. Wikelski. 2008. Going, going, gone: Is animal migration 
disappearing? PLoS Biology 6:1361?1364. 
Winker, K., J. Rappole, and M. Ramos. 1995. The use of movement data as an assay 
of habitat quality. Oecologia:211?216. 
164 
 
 
  
Wolda, H. 1978. Seasonal fluctuations in rainfall, food and abundance of tropical 
insects. The Journal of Animal Ecology 47:369?381. 
Yong, W., and F. R. Moore. 2005. Long-distance bird migrants adjust their foraging 
behavior in relation to energy stores. Acta Zoologica Sinica 51:12?23. 
Zahavi, A. 1971. The social beahviour of the White Wagtail Motacilla alba alba 
wintering in Israel. Ibis 113:203?211. 
Zuckerberg, B., D. Fink, F. A. La Sorte, W. M. Hochachka, and S. Kelling. 2016. 
Novel seasonal land cover associations for eastern North American forest birds 
identified through dynamic species distribution modelling. Diversity and 
Distributions 22:717?730. 
Z??iga, D., J. Falconer, A. M. Fudickar, W. Jensen, A. Schmidt, M. Wikelski, and J. 
Partecke. 2016. Abrupt switch to migratory night flight in a wild migratory 
songbird. Scientific Reports 6:34207. 
Zvoleff, A. 2015. gfcanalysis: Tools for Working with Hansen et al. Global Forest 
Change Dataset. https://CRAN.R-project.org/package=gfcanalysis. 
 
 
165