Population Structure in Western Sandpipers (Calidris Mauri): Variation in Genes, Morphology and Vocalizations in a Migratory Shorebird

Home , Common sandpiper, Least sandpiper, Pectoral sandpiper, Purple sandpiper, Stint

Population structure in Western Sandpipers (Calidris mauri): variation in genes, morphology and vocalizations in a migratory shorebird

by Birgit Schwarz Diplom, University of Tübingen, 2007

Thesis Submitted in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

in the Department of Biological Sciences Faculty of Science

 Birgit Schwarz 2016 SIMON FRASER UNIVERSITY Spring 2016

Approval

Name: Birgit Schwarz Degree: Doctor of Philosophy Title: Population structure in Western Sandpipers (Calidris mauri): variation in genes, morphology and vocalizations in a migratory shorebird Examining Committee: Chair: Gordon Rintoul Associate Professor

Ronald Ydenberg Senior Supervisor Professor

David Lank Co-Supervisor Adjunct Professor

Felix Breden Supervisor Professor

Darren Irwin Supervisor Professor Department of Zoology University of British Columbia

John Reynolds Internal Examiner Professor

Elizabeth MacDougall-Shackleton External Examiner Associate Professor Department of Biology Western University

Date Defended/Approved: January 25, 2016

Ethics Statement

iii

Abstract

Effective conservation management of migratory birds can be challenging and requires knowledge about population structure and the strength of migratory connectivity. The latter likely affects the degree to which populations are locally adapted and differentiated, and both may affect their ability to adapt to environmental changes.

Western Sandpipers (Calidris mauri) are long-distance migrants that exhibit latitudinal differences in morphology and life-history strategies across their non-breeding range. The differential migration patterns in this small shorebird species could be based on population genetic differences or phenotypic plasticity. I investigated the population structure of Western Sandpipers, and its implications for differential migration and conservation, across their global range using genetic, morphometric and acoustic data. I recorded and analyzed breeding male vocalizations, conducted playback experiments and collected genetic and morphometric data across breeding and non-breeding ranges.

Amplified Fragment Length Polymorphisms (AFLP) and mtDNA sequence data indicated low, but significant genetic population structure across breeding and non-breeding ranges. I found evidence for an evolutionarily recent demographic expansion (mtDNA), for latitudinal clines in genetic scores (AFLP) and for isolation by distance (AFLP). A scan of thousands of AFLP markers in a pooled lane approach revealed no fixed genetic differences among juveniles with a postulated difference in life-history strategy or migratory direction. Songs, but not alarm calls, varied geographically, decreasing in length and increasing in fundamental frequencies from southern to northern breeding sites. These geographic differences were sufficiently large to allow males to discriminate between local and non-local songs and hence are potentially biologically relevant. AFLP scores and vocalization frequencies showed correlations with body size. In conjunction, my results suggest that Western Sandpipers underwent a recent range and population expansion that has resulted in divergence patterns that are primarily isolation by distance driven. My results suggest interrelationships among genetic population structure, morphology and latitudinal segregation. While populations were not sufficiently distinct to allow for assignment tests, the latitudinal gradients found across both the breeding and non-breeding grounds are suggestive of a gradual ’chain’-like migration pattern, but do not indicate strong migratory connectivity. Consequently, conservation strategies could focus on the protection of major sites at different latitudes.

Keywords: population genetics; shorebird; migration; vocalizations; bird song; geographic variation

Dedicated to:

~ Ian Jong ~ partner, friend, love of my life

~ Petra Schwarz ~ sister at heart

~ Björn Siemers ~ inspiring scientist and mentor extraordinaire

Acknowledgements

First and foremost, I would like to thank my supervisor, David Lank, for taking me on as a student and guiding me on this journey towards my PhD. Thank you for being so readily available whenever I had a question or needed help or guidance and at the same time giving me enough freedom to make this project my own and pursue the questions I was interested in. I truly appreciate your patience and enthusiasm, the many insightful discussions over the years, and how generous you have been with your time and in sharing your considerable expertise.

Secondly, I would like to express my gratitude to the other members of my supervisory committee for their guidance and feedback: Ron Ydenberg, Felix Breden and especially to Darren Irwin for sharing his molecular and song expertise and for letting me work in his lab. I would also like to thank David Green, Beth MacDougall-Shackleton and John Reynolds for their helpful feedback and suggestions, and Kathleen Fitzpatrick for being such a great mentor.

I am deeply indebted to a large number of colleagues, collaborators and field assistants who generously shared existing data, helped collect and analyze the enormous amount of data for this project and/or provided their expertise. I am truly honoured that I got the chance to work with so many amazing scientists. I would especially like to thank Samantha Franks, Rick Lanctot, Guillermo Fernández, Diane Tracy, Brett Sandercock, River Gates, Alberto Castillo, Mark Colwell, Jorge Correa, Alexey Dondua, Ben Haase, Brooke Hill, Annie Schultz, Jay Schamel, Dan Ruthrauff, Eunbie Kwon, Dave Hope, Ummat Somjee, David Hodkinson, Gillian Andersen, Nathan Hentze and Toby St. Clair. Thank you also to Luis Enríquez-Paredes for sharing primers and molecular expertise, the late Alan Baker for fruitful discussions of molecular analyses and Dave Toews and Alan Brelsford for their invaluable help with the lab work and their advice regarding data analyses. Dave also provided songbird recordings to serve as controls in my playback experiments.

The shorebird research community, especially the Western Hemisphere Research Group and its meetings, has been a great source of inspiration and I am very grateful to the many people who provided ideas and valuable feedback over the years. The Arctic Shorebird Demographics Network was instrumental in facilitating collaborations and data sharing.

Financial support was provided by the Natural Sciences and Engineering Research Council Strategic Grant and Discovery grant programs, the Neotropical Migratory Bird Conservation Act grant program administered by the US Fish and Wildlife Service, the Alaska Department of Fish and Game, the National Science Foundation’s Arctic Natural Sciences program, Environment Canada, the Mexican National Council for Science and Technology, Universidad Nacional Autónoma de México, the DAAD (German Academic Exchange Service), the Daimler und Benz Stiftung, the Centre for Wildlife Ecology and Simon Fraser University.

The Sitnasuak, Wales and Kikiktagruk Native Corporations, Kotzebue Electric Association Cooperative (especially Brad Reeve), the Ecuasal Company and the various refuges in Mexico, Florida, Puerto Rico, South Carolina, Texas, and the YK-Delta supported our research by generously allowing us access to their lands. The Smithsonian Tropical Research Institute in Panama provided logistical support.

I am very grateful to Tina Moran, Randy Meyers and Jim Dau for their hospitality, local expertise, logistical support and assistance with the field work – your enthusiastic help made our field work at the Kotzebue site not only possible but also made our stay a very pleasant experience. I was truly amazed by the warm welcome and generous hospitality that we received in Nome. Pearl Johnson, Dawyn Sawyer, and Jim Menard – thank you so very much for going above and beyond in making our stay as safe and enjoyable as possible. Thank you also to the staff of the Alaska Department of Fish

vii

and Game (especially Peter Bente and Gay Sheffield) and the National Park Service for providing advice and assistance with logistics.

Connie Smith, Monica Court, Marlene Nguyen and Debbie Sandher provided administrative support, lovely conversations and were always willing to lend a helping hand. Connie Smith has been the heart and soul of the CWE since before I joined the lab and her administrative support has been absolutely invaluable.

I am profoundly grateful to all the current and former members of the CWE and Irwin labs who provided so much advice, feedback, support and friendship over the years – they are the people that have made this journey truly special. I would especially like to express my gratitude to: Dave Hope, David Hodkinson, Anna Drake, Simon Valdez, Tony St. Clair, Willow English, Eve Tavera Fernandez, Cailin Xu, Richard Johnston, Philina English, Marinde Out, Sarah Jamieson, Ariam Jiménez, Christine Rock, Michaela Martin, Rachel Gardiner, Dave Toews, Alan Brelsford, Kira Delmore, Julie Lee-Yaw and Miguel Alcaide.

Out of all of these colleagues and friends, one stands out in particular. Samantha Franks, as you wrote so eloquently in your own acknowledgements, you were “the best person I could ever have hoped to work with during a gruelling four-month field season travelling around the US and Latin America capturing sandpipers”. Sharing all these adventures with you was truly special and I cherish the friendship that grew out of these adventures and continues to be a source of great joy to me. Thank you for your invaluable help, from sharing your knowledge about sandpipers and how to catch them to help with R and intellectual discussions, you have had a crucial hand in making this thesis possible.

I also wish to thank my family and friends for their unwavering support. I would like to particularly thank Regina Schwarz and Susanne Mulert for their amazing friendship over all these years. Thank you also to the Jong family (Elgie, Tony and Sumi) for all the laughter, support and your role in keeping me happy and healthy. My family

viii

has been a great source of support in numerous different ways. Thank you to my parents and grandparents, and especially to Petra Schwarz. As the only other academic in the family you have in many ways paved the way for me. Thank you for being such an amazing source of humour and support over all these years. Not to mention the very direct effect you have had – on my academic progress in general and on the successful completion of this thesis specifically – by providing me with the all those books and articles over the years (especially the hard to find ones). I could not wish for a better sister.

My most profound gratitude goes to my partner Ian. Where to begin? I cannot even list the countless ways in which you have been instrumental for the completion of my PhD. Thank you most of all for all the love and emotional support you have given me and continue to give. You made all the fun parts even more enjoyable and considerably eased those rough patches along the way. I am also incredibly grateful for your amazing practical and intellectual help: from fixing electronic equipment with a pocket knife in the middle of the tundra, to helping with figures and providing fantastic editorial support and insightful discussions, this thesis just simply would not exist without you. Thank you for taking care of me and for believing that I could do this.

Table of Contents

Approval ...... ii Ethics Statement...... iii Abstract ...... iv Dedication ...... v Acknowledgements ...... vi Table of Contents ...... x List of Tables ...... xiii List of Figures ...... xvii Introductory Image...... xxvi

Chapter 1. Population structure in Western Sandpipers (Calidris mauri): variation in genes, morphology and vocalizations in a migratory shorebird - Overview and synthesis ...... 1 1.1. Introduction ...... 1 1.1.1. Bird migration, population structure and conservation ...... 1 1.1.2. Migratory shorebirds ...... 2 1.1.3. Western Sandpipers ...... 4 Chapter 2: ...... 6 Chapter 3: ...... 6 Chapter 4: ...... 6 Chapter 5: ...... 7 Chapter 6: ...... 7 1.2. Synthesis of the main thesis results, their implications and suggestions for future research ...... 7 1.2.1. Population structure in Western Sandpipers ...... 7 1.2.2. Implications for migratory connectivity and differential migration ...... 11 1.2.3. Conservation implications and future studies ...... 13 1.3. References ...... 15

Chapter 2. Population Genetics of Western Sandpipers ...... 22 2.1. Abstract ...... 22 2.2. Introduction ...... 23 2.3. Methods...... 27 2.3.1. Sampling and DNA extraction ...... 27 2.3.2. mtDNA sequencing ...... 29 2.3.3. AFLP genotyping ...... 29 2.3.4. Morphometric data collection: ...... 30 2.3.5. Data analysis ...... 31 mtDNA: ...... 31 AFLP: ...... 32 Correspondence of Morphology and AFLP data: ...... 35 2.4. Results ...... 35

2.4.1. mtDNA ...... 36 2.4.2. AFLP: ...... 37 2.4.3. Correspondence of Morphology and AFLP data: ...... 39 2.5. Discussion ...... 40 2.5.1. mtDNA ...... 40 2.5.2. AFLP: ...... 46 2.5.3. Correspondence of Morphology and AFLP data: ...... 48 2.6. Synthesis ...... 52 2.7. References ...... 55 2.8. Tables ...... 68 2.9. Figures...... 75 2.10. Appendix ...... 85

Chapter 3. Songs differ but alarm calls do not: a study of geographical variation in Western Sandpipers ...... 94 3.1. Abstract ...... 94 3.2. Introduction ...... 95 3.3. Methods...... 98 3.3.1. Spectrographic Analysis: ...... 99 3.3.2. Song: ...... 99 3.3.3. Alarm calls: ...... 100 3.3.4. Statistical tests ...... 100 3.4. Results ...... 101 3.4.1. Song ...... 101 3.4.2. Alarm calls ...... 102 3.5. Discussion ...... 102 3.6. References ...... 108 3.7. Tables ...... 116 3.8. Figures...... 118 3.9. Appendix ...... 125

Chapter 4. Inter- and intra-individual variation in the vocalizations of male Western Sandpipers ...... 131 4.1. Abstract ...... 131 4.2. Introduction ...... 132 4.3. Methods...... 137 4.3.1. Vocal individuality – evidence for identity signalling: ...... 138 4.3.2. Correlation of body size and vocal frequency characteristics: ...... 139 4.3.3. Reproductive status and song characteristics: ...... 140 4.4. Results ...... 141 4.4.1. Vocal individuality – evidence for identity signalling: ...... 141 4.4.2. Correlation of body size and vocal frequency characteristics: ...... 141 4.4.3. Reproductive status and song characteristics: ...... 142

4.5. Discussion ...... 142 4.5.1. Vocal individuality – evidence for identity signalling: ...... 142 4.5.2. Correlation of body size and vocal frequency characteristics: ...... 146 4.5.3. Reproductive status and song characteristics: ...... 150 4.6. Conclusions ...... 151 4.7. References ...... 151 4.8. Figures...... 171 4.9. Appendix ...... 175

Chapter 5. Does song function in territorial defense in the Western Sandpiper? An experimental approach...... 181 5.1. Abstract ...... 181 5.2. Introduction ...... 182 5.3. Methods...... 184 5.3.1. Behavioural observations: ...... 184 5.3.2. Playback experiments: ...... 184 Subjects ...... 184 Stimuli ...... 185 Playback procedure and data collected ...... 185 Data collection and statistical analysis: ...... 186 5.4. Results ...... 187 5.4.1. Behavioural observations: ...... 187 5.4.2. Playback experiments: ...... 187 5.5. Discussion ...... 188 5.6. References ...... 189 5.7. Tables ...... 195 5.8. Figures...... 196

Chapter 6. Do male Western Sandpipers discriminate among local and non-local songs? ...... 198 6.1. Abstract ...... 198 6.2. Introduction ...... 199 6.3. Methods...... 201 6.3.1. Subjects and study site ...... 201 6.3.2. Stimuli ...... 201 6.3.3. Playback procedure ...... 202 6.3.4. Data collection and statistical analysis: ...... 202 6.4. Results ...... 204 6.5. Discussion ...... 204 6.6. References ...... 207 6.7. Figures...... 214 6.8. Appendix ...... 215

xii

List of Tables

Table 2-1. Sample sizes. Total numbers of samples that were included in mtDNA, AFLP and morphometric analyses respectively. Site-specific sample sizes are listed by sex and for non-breeding sites by age group. Numbers in brackets for the morphometric data refer to individuals for which both genetic and complete morphological data was available and which were therefore included in the individual-based analyses comparing morphological and AFLP data...... 68 Table 2-2. Occurrence of each haplotype. Each row lists the total number of individuals with the respective haplotype in the sample as a whole and across the different sampling sites. For non-breeding sites, numbers are listed separately for each sex. The right-most column indicates corresponding haplotypes found by Enríquez-Paredes et al. (2012)...... 69

Table 2-3. FST values across the sampled breeding and non-breeding range for mitochondrial control region sequence and AFLP data. For the nuclear AFLP data, all three FST analogues indicated low, but significant levels of genetic divergence across both the breeding and non-breeding range. MtDNA haplotype frequency based measures suggested a moderate level of divergence. No evidence for lineage divergence was found, however: molecular distance based measures did not differ significantly from 0 for either the breeding or the non-breeding range...... 71

Table 2-4. Population pair-wise FST values for the mtDNA control region sequence (681bp). As males and females may differ in migratory strategies, they were entered separately into the analysis for all non-breeding sites. FST values were calculated in a haplotype frequency based approach using the AMOVA function in Arlequin. FST results above the diagonal were obtained by including all haplotypes in the analysis. Values below the diagonal represent the analysis for which all singleton haplotypes were excluded. Bold FST values are significant at P, q and lFDR<0.01. Underlined pair-wise comparisons were considered significant in one but not the other of the two analyses...... 72

xiii

Table 2-5. AFLP Primer combinations and number of polymorphic bands. Combinations of primers in the AFLP analysis that yielded consistent and informative band patterns and the resulting number of reliable, polymorphic markers included in analyses. All primer combinations were used for the analysis of breeding ground samples. CAT primer combinations were not run for non-breeding ground samples...... 73 Table 2-6. Population pair-wise comparisons of AFLP band patterns. Values above the diagonal represent Cavalli-Sforza and Edwards chord distances (Takezaki formula). These distances were employed as a measure of genetic distance in isolation by distance analyses and calculated within, but not between breeding and non-breeding ranges. Pairwise FST values are given below the diagonal and were derived from an analysis of band frequency based distance measure (Dice distances). As males and females may differ in migratory strategies, they were entered separately into the analysis for all non-breeding sites. FST values that are significant at P, q and lFDR < 0.05 are given in bold...... 74 Table 2-7. Haplotype definitions. Only positions that were variable in our sample are listed, with position 1 corresponding to the first nucleotide of the sequence analyzed in this study. The last column lists corresponding haplotypes found by Enríquez-Paredes et al. (2012)...... 88 Table 2-8. Relative frequency of occurrence of each haplotype. The overall frequency within the sample and relative occurrence within each site are given in percent. For non-breeding sites, percentages are listed separately for each sex. The right-most column indicates corresponding haplotypes found by Enríquez-Paredes et al. (2012)...... 90 Table 2-9. AFLP primer pairs run in pooled approach. Each of the listed primer combinations marked with an X were surveyed for fixed genetic differences among juvenile females sampled at northern, southern and eastern non-breeding sites. No fixed differences were detected for any of the thousands of possible markers. Marker combinations denoted by an asterisk were run in the single-lane approach, a hyphen indicates un-tested combinations...... 92 Table 2-10. Component loadings for Morphology PC1. All morphometric measurements loaded strongly and positively onto the first PC, suggesting that this component reflects overall size...... 93 Table 3-1. Means for selected frequency and temporal variables of male songs and alarm calls: Means and standard deviations for each site were calculated from individual male means and are presented for descriptive purposes. We restricted our analysis to the fundamental for the frequency domain...... 116

xiv

Table 3-2. Mean coefficients of variation (COVs) for each variable type: Mean COVs were calculated from the COVs of all variables of each type: 10-13 for songs, 4 for alarm calls. None of the mean COVs for alarm calls are lower than those for song. Alarm call features are therefore either equally variable as or more variable than song features...... 117 Table 3-3. Mean values of all song and alarm call variables that were entered into the PCA. Means and standard deviations are given for each breeding site. Within each selection the variable Peak Frequency refers to the frequency with the highest power, Center Frequency to the frequency at which the selection can be divided into two sections with equal amounts of energy, and 90% Bandwidth describes the frequency interval that encompasses 90% of the energy, with 5% of the energy above and below this interval. The song selections i1, i2, i3 and i4 refer to the first, second, third and last introductory note respectively, t1 encompasses the first trill pulse, t2 pulses 1-5, t3 pulses 6-10, t4, t5 and t6 the third-last, second-last and last pulse group of the trill and t7 the last 12 contiguous pulses (see Fig. 3-1). The alarm call selection 1 encompasses the entire call, whereas selections 2, 3 and 4 contain the first, third and last note respectively (see Fig. 3-2)...... 126 Table 3-4. Component loadings for the first three principal components for songs and alarm calls. Variables that loaded highest onto PC1 are listed first. Loadings above 0.550 for each component are highlighted in bold, as are the corresponding variables for PC1...... 129 Table 4-1. Component loadings for the Principal Components (PCs) used in Discriminant Function Analyses (DFA) of songs (a.) and alarm calls (b.). Only PCs with eigenvalues > 1 entered the DFA and are listed here...... 175 Table 4-2. Contributions of the Principal Component variables (PCs) to each Discriminant Function for songs (a.) and alarm calls (b.). Only PCs that entered the analysis are shown and are ordered by the absolute size of their correlation within the function...... 178 Table 4-3. Component loadings for the first Principal Component (song frequency PC1) of the song frequency Principal Component Analysis for males in the alarm call (a.) and morphology (b.) correlation analyses. Following a parallel analysis, solely the first Principal Component was retained for further analysis; variables are listed in order of the highest loadings onto PC1. For details on selections and variables see chapter 3...... 179

Table 4-4. Component loadings for the first Principal Component of the morphology Principal Component Analysis. As all morphological measurements loaded positively onto the first Principal Component, it was retained as a measure of body size. Variables are listed in order of their loadings onto PC1 with higher-loading variables listed first...... 179 Table 4-5. Correlations of univariate song frequency measures with a body size index (morphology PC1) for males from all sites. See chapter 3 for a detailed description and visualization of the measured frequency variables. Significant correlations are bolded and effect sizes are listed for significant correlations only. Data are presented primarily for descriptive purposes to evaluate the relative effect of body size on the frequencies of different song components, following a significant multivariate effect; consequently we did not apply P- value corrections for multiple comparisons...... 180 Table 5-1. Percentage of playback songs that were overlapped by male song responses. In 66.7% of the 24 trials males overlapped with one or more playback songs by starting to sing prior to the end of the playback songs...... 195 Table 5-2. Occurrence of agonistic behaviour or direct approaches in response to song playbacks. Agonistic behaviour included wing-up displays, upright postures, crouched posture approaches, approaches with chase note vocalizations, physically contacting or chasing other birds. Pre-incubation males (N=12) frequently approached the decoy and displayed agonistic behaviour, incubating males (N=12) rarely did...... 195 Table 6-1. Component loadings for song response PC1 (a.) and approach and agonistic behaviour PC1 (b.). Based on these loadings both PC1s represented a measure of the strength of the respective response and were retained for further analysis...... 215 Table 6-2. Summary statistics for univariate response variables. Mean, Standard Deviation (SD), Median, 1st and 3rd Quartile (Q1, Q3) values for each of the measured response variables for the three playback types (n=18)...... 216

xvi

List of Figures

Figure 2-1. Map depicting sampling site locations. Blood samples for genetic analyses were obtained from five breeding sites (Siberia, Barrow, Kotzebue, Nome and YK-Delta) and eight non-breeding sites. The latter effectively represented six geographic locations (California, Western Mexico, Yucatan, Puerto Rico, Panama and Ecuador)...... 75 Figure 2-2. Median-joining haplotype network. Haplotypes are denoted by circles, each connecting line corresponds to one mutation. Circle size is proportional to the number of individuals with this haplotype and colors indicate the subsample(s) in which each haplotype was found. Small black circles designate presumed intermediate haplotypes not present in the sampled individuals. The central, most common haplotype (Hap01) was found across all sites and was present in 45.8% of individuals. The remaining 39 haplotypes were relatively rare and differed in only 1-4 nucleotides from Hap01, resulting in a star-shaped network pattern that may indicate recent demographic expansion...... 76 Figure 2-3. Distribution of haplotypes across breeding sites. The four circles show the haplotype composition at each of the four breeding sites. Circle sizes are proportionate to sample sizes. Each circle segment denotes a different haplotype with segment size indicating the proportion of individuals that possess the respective haplotype. Haplotypes that occurred at more than one site (including non- breeding sites) are shown in colour. White segments signify haplotypes that were found exclusively at the respective site. Hap01 was present at all sites, with a frequency of 32% among individuals at the Nome site and reaching 50% at the other three sites. Hap02 was shared among three, Hap03, 10, 12, 13 and 15 among two breeding sites. Hap04, 06, 07, 08, 09, 11 and 14 were found at only one breeding site, but were also present in individuals sampled on the non-breeding grounds (Fig. 2-4)...... 77

xvii

Figure 2-4. Distribution of haplotypes across non-breeding sites. The circles show haplotype compositions across different non-breeding sites, separated by sex where samples from both sexes were available (California and Panama). For California, data from two sampling sites were pooled. Circle sizes are proportionate to sample sizes. Each circle segment denotes a different haplotype with segment size indicating the proportion of individuals that possess the respective haplotype. Haplotypes that occurred at more than one site (including breeding sites) are shown in colour. White segments signify haplotypes that were found exclusively at the respective site. Hap01 was present at all sites, its prevalence ranging from 33- 60% for each subsample. Hap02, 03, 04, 05 were shared among two non-breeding sites. Hap 06, 07, 08, 09, 11, 14 were found at only one non-breeding site, but were also present in individuals sampled on the breeding grounds (Fig. 2-3) ...... 78 Figure 2-5.Correlation between genetic and geographic distances. Mantel tests confirmed a significant positive correlation between genetic and geographic distance among both the breeding (a) and the non- breeding (b) sites (Z=299, r=0.582, P=0.032, and Z=1721, r=0.620, P=0.022, respectively). Geographic distance accounted for 34% of the variation in genetic distance among breeding sites, strongly suggesting an isolation-by-distance pattern. This pattern was mirrored across the non-breeding sites where geographic distance accounted for 39% of the variation in genetic distance...... 79

xviii

Figure 2-6. Geographical representation of the first three spatial Principal Component scores on the breeding grounds. Individual spatial Principal Component (sPC) scores, derived from AFLP profiles, are shown on a map to visualize spatial relationships. The analysis is based on a modified Principal Component Analysis that explicitly takes spatial information into account. Individuals sampled at five breeding sites are represented by squares. Square color and size reflect individual sPC scores: white squares correspond to negative, black squares to positive values and size indicates magnitude. Consequently, a large white square vs. a large black square represents a greater genetic dissimilarity than a large vs. a small white square or a small white vs. a small black square. To achieve a better graphical representation, points were randomly scattered around the respective sampling sites and hence do not reflect exact sampling locations. The inset eigenvalues barplot depicts the magnitude of the eigenvalues of each score – positive values for global scores and negative for local scores. The respective eigenvalues for the scores depicted in each map are marked black. a. Increasing sPC1 scores from south to north indicate a latitudinal cline across the Alaskan breeding sites. The pattern is suggestive of isolation by distance. b. sPC2, on the other hand, appears to highlight differences between the easternmost breeding site in Siberia and the Alaskan breeding sites, with the geographically closer sites (Kotzebue and Nome) appearing less dissimilar than the more distant ones (YK-Delta and Kotzebue). c. The last retained global score (sPC3) appears to differentiate the centrally located Nome and, to a lesser degree, the Kotzebue site from the peripheral breeding sites...... 81

xix

Figure 2-7. Geographical representation of the first spatial Principal Component scores on the non-breeding grounds. Individual spatial Principal Component (sPC) scores, derived from AFLP profiles, are shown on a map to visualize spatial relationships. The analysis is based on a modified Principal Component Analysis that explicitly takes spatial information into account. Individuals sampled across six non-breeding areas are represented by squares. Square color and size reflect individual sPC scores (see Fig. 6 for a detailed explanation). To achieve a better graphical representation, points were randomly scattered around the respective sampling sites and hence do not reflect exact sampling locations. The inset eigenvalues barplot depicts the magnitude of the eigenvalues of each score – positive values for global scores and negative for local scores. Marked black is the eigenvalue for the first score depicted in the map. Non-breeding scores are based on separate analysis and are therefore not directly comparable with breeding ground scores. Similarly to the breeding ground data, the sPC1 scores decrease from north to south, suggesting a gradual isolation by distance pattern...... 82 Figure 2-8. Correlation between individual body size and genetic scores for females and males. The morphometric PC1 is based on culmen, tarsus and wing length, reflecting body size. Genetic scores are derived from a spatial PCA, reflecting the sPC1s for breeding and non-breeding samples respectively. For breeding females no relationship between body size and genetic scores was detected (r=0.036, P=0.411, n=41), whereas a moderately strong, negative correlation was present among males (r=-0.356, P=0.018, n=35). Across the non-breeding range, both sexes showed significant, inverse correlations among body size and genetic scores (r=-0.280, P=0.019, n=55 and r=-0.344, P=0.036, n=28)...... 83

Figure 2-9. Average genetic scores and latitude (left); correlation between average body size and genetic scores for females (center) and males (right). The datapoints represent mean scores of the first principal components (PC) for each breeding or non-breeding site. The first genetic PC is based on the analysis of a reduced set of 22 AFLP markers that were identified as spatially informative on the non- breeding grounds through the spatial PCA. Left: mean AFLP PC1 (both sexes; error bars=95% C.I.) latitudinal trends have the same directionality for breeding and non-breeding scores. Center and right: the first morphometric PC is based on culmen, tarsus and wing measurements and appears to reflect body size (positive loadings of all three measurement types). Mean morphometric and genetic PC1 scores are strongly correlated for both sexes across sites (females: r=0.650, P=0.021, n=10 sites; males: r=0.727, P=0.013, n=9 sites)...... 84 Figure 2-10. Eigenvalues plots for breeding (a.) and non-breeding (b.) range spatial Principal Component Analyses. The screeplots depict the decomposition of the eigenvalues into their variance and autocorrelation components. The vertical, dashed line on the right represents the largest achievable variance through a linear combination of AFLP markers (= that of the first PC if a regular PCA is conducted on these markers), whereas the dashed horizontal lines at the top and bottom depict the respective limits for Moran’s I. For the breeding sites sPCA, the first three components contain the highest degree of variability and spatial structuring, clearly stood out from the remaining eigenvalues and were therefore retained after global analysis indicated significant global structure. For the non-breeding sites sPCA only the first component contained sufficient spatial and variance information to be interpreted after a significant global test...... 85 Figure 2-11. AFLP sPC scores for breeding individuals. The first sPC is shown in relation to the second (left graph) and third (right graph) sPCs. While individuals cluster by site, a substantial degree of overlap exists, particularly among neighbouring sites. SPC1 scores show a gradual increase from the southernmost (YK-Delta) to the northernmost (Barrow) sampling site, indicating a cline across the Alaskan breeding sites. SPC2 and 3 accentuate the easternmost (Siberia) and central (Nome) breeding sites respectively...... 86 Figure 2-12. AFLP sPC1 scores for non-breeding individuals in relation to latitude for each sex and age group. SPC1 scores show a consistent increase with latitude across all of the four sex and age groups...... 87

xxi

Figure 3-1. Male Western Sandpiper song spectrograms. The spectrogram in part A depicts the whole song, part B the trill in greater detail, with the oscillogram in the grey upper panel in B clearly showing the pulsed nature of the trill. Spectrograms were generated in Raven 1.4 (Hann window, window size of 312 points with 90% overlap, grid spacing 93.8Hz, 3dB filter bandwidth 221Hz). We removed low-frequency noise and cut spectrograms off at 12.75kHz solely for visualization purposes. We only analyzed the fundamental frequency in the frequency domain, labels indicate song components selected for analysis, BW=bandwidth. For a detailed list and description of measured variables see Table 3-3 in the appendix...... 118 Figure 3-2. Male Western Sandpiper alarm call spectrogram. The spectrogram was generated in Raven 1.4 (Hann window, window size of 312 points with 90% overlap, grid spacing 93.8Hz, 3dB filter bandwidth 221Hz). We removed low-frequency noise and cut spectrograms off at 12.75kHz solely for visualization purposes. We only analyzed the fundamental in the frequency domain, labels indicate alarm call components selected for analysis. For a detailed list and description of measured variables see Table 3-3 in the appendix...... 119 Figure 3-3. Spectrograms depicting three male songs (A) and alarm calls (B) from each of the four breeding sites. Spectrograms were generated in Raven 1.4 (Hann window, window size of 312 points with 90% overlap, grid spacing 93.8Hz). We removed noise below 1kHz and cut spectrograms off at 12.75kHz solely for visualization purposes. The examples were chosen to reflect some intra-site variation and are not necessarily representative in every aspect. The first two alarm calls at the Wales site were recorded from the same individual, all other alarm calls and all songs represent different individuals...... 120 Figure 3-4. Individual male principal component scores for song: Scatterplot depicting the first two principal component scores for each male. Our spectrographic analysis yielded 46 frequency and temporal variables which were entered into the PCA. Between 3 and 15 songs were analyzed per male, but only the mean values for each male were entered into the analysis. Symbols represent sites at which males were recorded...... 121

xxii

Figure 3-5. Mean PC1 scores for male song at the different breeding sites: Error bars represent 95% Confidence Intervals. Male songs at the southernmost site in the YK-Delta differed significantly from songs at the other three sites (F = 15.63, df = 3 and 41, P < 0.001; Tukey post- hoc test significant at P<0.001, <0.01 and <0.001 for YK-Delta vs. Nome, Wales and Kotzebue and P<0.05 for Wales vs. Kotzebue). Songs recorded at Wales and Kotzebue also differed from each other...... 122 Figure 3-6. Individual male principal component scores for alarm calls: Scatterplot depicting the first two principal component scores for each male. Our spectrographic analysis yielded 17 frequency and temporal variables which were entered into the PCA. Between 5 and 9 alarm calls were analyzed per male, but only the mean values for each male were entered into the analysis. Symbols represent sites at which males were recorded. Alarm calls from all sites overlap and no separation among sites is evident...... 123 Figure 3-7. Mean PC1 scores for male alarm calls at the different breeding sites: Error bars represent 95% Confidence Intervals. Alarm calls did not differ among sites...... 124 Figure 4-1. Scatterplot of discriminant function 1 (DF1) and 2 (DF2) values for male songs (a) and alarm calls (b). Each datapoint in the scatterplot represents a single song (a) or alarm call (b). Individual males (n=13 (a) and n=7 (b)) are represented by different symbols, their centroids through numbered squares. Frequency and temporal variables both entered into a Principal Component Analysis. We included all PCs with eigenvalues > 1 in a stepwise DFA model (threshold of F=3.84 for entry and F=2.71 for removal of a variable; separate groups covariance matrix; prior probabilities based on sample size). For songs (a) the DFA assigned 92.8% of songs to the correct males based on the first seven discriminant functions, with DF 1 and 2 accounting for 60.2 % of the variation. All alarm calls (b) were correctly assigned using the first five discriminant functions, DF1 and 2 accounted for 77.4% of the variation...... 171 Figure 4-2. Correlation of alarm call center frequency with song frequency PC1: Each datapoint represents one male (n=21) and data from all four breeding sites were entered into the analysis. The variation in alarm call frequency explained 34.5% of the variation in the first song frequency Principal Component (P=0.003), revealing a strongly positive relationship between the frequencies of the two vocalization types...... 172

xxiii

Figure 4-3. Correlation of body size with song (a) and alarm call frequencies (b) across breeding sites. Morphology PC1, a proxy of body size, is based on culmen, tarsus, wing and mass measurements, accounting for 38% of the total variation. a: Song frequency PC1, which accounted for 65.9% of the variation in song frequencies, showed a moderately strong negative relationship with body size (n=22, r=- 0.372, P=0.044). b: Alarm call center frequency showed a somewhat weaker negative correlation with body size that approached significance (n=25, r=-0.32, P=0.059)...... 173 Figure 4-4. Effect of reproductive state on song characteristics. We conducted a PCA on the song parameters of pre-incubation, incubating and brood-rearing males at the Nome breeding site. While reproductive stage did not appear to affect PC1 and 2 (not shown), PC3 differed among pre-incubation and chick-rearing males (n=30, F=5.5, df=2 and 27, P=0.01). PC3 accounted for 10.3% of the total variation in song characteristics and primarily represented the number of introductory notes, their duration (inverse loadings) and bandwidth characteristics of terminal trill components. Error bars represent 95% confidence intervals...... 174 Figure 5-1. Photograph of the wooden decoy used in the playback experiments. The decoy was placed on top of the loudspeaker during the experiments...... 196 Figure 5-2. Response of male Western Sandpipers to song playback. Responses are shown for pre-incubation and incubating males. Each playback stimulus consisted of three songs and each male was subjected to both treatments: Western Sandpiper song and control playbacks (Black-throated Green Warbler song). Boxes depict the interquartile range (IQR) with the median line, whiskers the range, with outliers denoted as stars (>3xIQR). A.: Number of songs given by males in response to playback; the differential was obtained by subtracting two times the number of songs given during the 30s pre-playback period from the number of songs given during the 60s post-playback period. Males responded to Western Sandpiper song, but not to controls. The median number of songs in incubating males was lower than in pre-incubation males but not significantly so. B. Latency (s) between the onset of the first playback song and the first song response. If no response occurred, the value was set to 60s. Males typically responded very quickly to playbacks of Western Sandpiper song...... 197

xxiv

Figure 6-1. Song (a.) and approach/agonistic behaviour (b.) responses to local and non-local song playbacks. PCAs were used to create a composite measure for the song response (a.) of male subjects and a separate measure for their approach and agonistic behaviour response (b.). In both cases, PC1 represented a measure of response strength and was therefore retained. The boxplots depict the responses to three different playback treatments: local songs recorded at the Nome site and non-local songs recorded at breeding sites near Kotzebue or in the YK-Delta, respectively (boxes depict the interquartile range (IQR) with the median line, whiskers the range, with outliers denoted as circles (>1.5xIQR) or stars (>3xIQR)). Each of the 18 male subjects was exposed to all three playback types. The strength of the song response differed among playback type with males responding more strongly to local than non-local song playbacks (Nome-Kotzebue: T=31, P=0.015, r=-0.508; Nome-YK- Delta: T=32, P=0.017, r=-0.497). Playback type did not, however, appear to affect approach and agonistic behaviour...... 214

xxv

xxvi

Chapter 1. Population structure in Western Sandpipers (Calidris mauri): variation in genes, morphology and vocalizations in a migratory shorebird - Overview and synthesis

1.1. Introduction

From their ability to fly to flashy plumages, from carefully choreographed mating display performances to marvellously complex, melodious songs and astonishing feats of migration, birds have long been a source of human fascination. Unsurprisingly, birds offer a cornucopia of study opportunities and have long been prime subjects for studying many aspects of evolution, ecology and behaviour. As is the case for so many taxa on our planet presently, conservation is increasingly a concern for many bird species. As a consequence, studying complex ecological and evolutionary relationships not only increases our understanding of biological processes and the natural world around us, but is also becoming a critical source of information for the development of effective conservation strategies.

1.1.1. Bird migration, population structure and conservation

Migratory species pose particular challenges for conservation management as their survival and reproduction success depends on multiple locations, i.e. not only on the availability of breeding habitat but also on non-breeding habitats that are spatially distinct from the breeding grounds, often thousands of kilometers away, as well as on habitats and conditions experienced en route (Marra et al. 1998; Webster et al. 2002; Faaborg et

al. 2010). Intraspecific variation in migration patterns and strategies among populations, sexes, age groups or individuals that differ e.g. in social status or body size further complicate the picture. Rather than being an exception the existence of such differential migration may actually be the ‘norm’ among migratory species (Cristol et al. 1999; Bell 2005). Where different segments of a population or different populations exhibit different spatial and/or temporal migration patterns, it is imperative for a successful conservation approach to take these patterns into account.

Conservation management of migratory bird species therefore benefits from knowledge of population structure and the strength of migratory connectivity between breeding and non-breeding sites (i.e. the extent to which individuals that originate from the same breeding population migrate to the same non-breeding site; Haig et al. 1997; Webster et al. 2002). The latter is likely to affect the degree to which populations are locally adapted (and thus divergence rates among populations; Webster and Marra 2005). The extent of genetic structuring and the strength of migratory connectivity may also affect the degree of flexibility with which populations may be able to adapt to changes (Webster and Marra 2005). However, information on migratory connectivity is scarce for many bird species, as it can be very challenging to keep track of migratory individuals throughout the annual cycle. Genetic analysis has the potential to be a powerful tool to reveal migratory connectivity patterns as well as providing crucial information about the population structure of a species. When supplemented with other data, e.g. on traits that may play a role in migratory behaviour or mate selection, it has the potential to help us gain a better understanding not only of the migratory patterns themselves, but also about what may be driving them, what effects partial habitat losses in specific areas are likely to have and how well individual species and populations are likely to be able to adapt to change.

1.1.2. Migratory shorebirds

Many shorebird species are long-distance migrants and the patchy distribution of many species on the non-breeding grounds along with their arctic breeding ranges make them particularly interesting subjects for studying the dynamics of migratory connectivity and population structure. Additionally, many, if not most of these species display at least some degree of differential migration and/or morphometric variation and what drives these patterns has not been adequately resolved at present.

Most shorebird species depend on coastal habitats or extensive wetlands and many are arctic or subarctic breeders and long-distance migrants. As such, they may be critically affected by coastal developments, reductions in wetland habitats and climatic changes. Multiple challenges and habitat loss across different parts of their annual cycle could have compounding effects. Indeed, apparent declines across many species are an increasing concern (Morrison et al. 1994; Morrison et al. 2001; Bart et al. 2007). However, changes in migratory behaviours may be affecting population estimates in some instances, potentially leading to biased numbers, incorrect conclusions and inefficient allocation of conservation resources (Ydenberg et al. 2004; Bart et al. 2007). Understanding migratory behaviour patterns and their underlying causes in shorebirds is therefore of prime importance. Flexibility in migratory behaviours via phenotypic plasticity may help to buffer shorebird populations to some degree in times of significant changes in the availability and location of suitable habitats. For arctic breeding species whose breeding habitats and migratory routes were particularly affected by glacial cycles, these cycles might have conferred a selective advantage to the presence of such flexibility. On the other hand, it has been suggested that the predictable and patchy distribution of coastal habitats may promote relatively rigid migratory routes (and hence strong migratory connectivity) for species that primarily used this kind of habitat. Furthermore, temporally separated migrations of adults and juveniles on their first migration in many species would appear to preclude parentally mediated learning of migratory routes and hence suggest that migratory behaviours and strategies may be innate to some degree.

1.1.3. Western Sandpipers

Western Sandpipers (Calidris mauri) are small, socially monogamous shorebirds that breed on the Arctic tundra within a relatively small geographic region in western Alaska and easternmost Siberia (Franks et al. 2014). They are long-distance migrants whose non-breeding habitats lie mainly along the Pacific coast of North America and range from British Columbia (Canada) to Peru, but to a lesser degree also include the Atlantic coast (New Jersey (U.S.) to Suriname) and the Caribbean (Franks et al. 2014). Adults and juveniles, as well as females and males exhibit differences in their migratory behaviour (Butler et al. 1987; Nebel et al. 2002). Adults depart the breeding grounds prior to their young, and within the non-breeding range, juveniles appear to be disproportionately represented near both the northern and southern edges (Butler et al. 1987; Nebel et al. 2002; Franks et al. 2014). Furthermore, females tend to abandon their broods earlier, and, on average, migrate further south than males (Nebel et al. 2002). Latitudinal clines in morphometric measures within each sex and apparent differences in life-history strategies among juveniles at different sites suggest that Western Sandpipers show differential migration across their non-breeding range even within sex and age classes. Specifically, average wing and culmen lengths appear to increase with decreasing latitude and juveniles that spend their first winter at northern sites appear to be more likely to return to the breeding grounds in their first year of life than those wintering further south, i.e. most juveniles further north on the non-breeding grounds were found to prepare for spring migration while hardly any juveniles did so at a southern site (Nebel 2005; O'Hara et al. 2005; O'Hara et al. 2006; Stein et al. 2008). These patterns suggest the possibility of distinct populations with different migration and life history strategies. We currently know very little about the genetic population structure of Western Sandpipers and how it might affect or may be affected by migration patterns. Furthermore, what little information is available is centered around a few sites on the non-breeding grounds and one breeding site. The respective studies suggest the presence of some degree of genetic differentiation across different non-breeding sites and/or habitats, but with their limited scope do not allow conclusions to be drawn regarding the

origin of the aforementioned morphological variation (Haig et al. 1997; Enríquez and Fernandez 2010; Enríquez-Paredes et al. 2012).

If there is genetic differentiation among Western Sandpiper populations, the question arises of how such differentiation might be maintained on the breeding grounds, given the species’ high mobility and limited breeding range. While adult return rates can be high, natal philopatry appears to be lower (ca. 3% return rate), at least on the spatial scale of study plots ranging from a few hectares to 4km2 in size (Holmes 1971, Sandercock 1997). As in many other bird species, male song could play a key role in this regard. Western Sandpiper song is a complex vocalization that consists of several introductory notes and a trill and appears to function in mate attraction (Brown 1962; Holmes 1973; Lanctot et al. 2000). Males emit songs during aerial display flights but may also sing on the ground (Brown 1962; Lanctot et al. 2000). As Western Sandpiper song appears to fulfill a mate attraction role, it likely is subject to sexual selection. As a sexually selected trait it not only has the potential to evolve more rapidly than other traits, but may also strongly affect gene flow (West-Eberhard 1983; Price 1998; Gonzalez- Voyer and Kolm 2011; Seddon et al. 2013). As such, a sexually selected trait such as song might be among the first traits to show geographical variation and simultaneously could be pivotal in promoting further differentiation (West-Eberhard 1983; Panhuis et al. 2001). As songs do not appear to be learned in shorebirds, genetic differentiation would be presumed to form the basis of any divergence in song characteristics (Miller 1996; Price 2007). Therefore, patterns of song variation might not only provide additional evidence for or against differentiation, but could – if such differentiation exists – also be a crucial mechanism in maintaining reproductive barriers.

The objective of this thesis project was to use genetic, morphometric and acoustic data to investigate multiple dimensions of the population structure of Western Sandpipers across the breeding and non-breeding range. Furthermore, I investigated whether genetic markers and morphological data could be used to establish patterns of migratory connectivity in this species. The questions that the individual chapters of this thesis will address are thus as follows:

Chapter 2:

In the first research chapter, I directly investigate the genetic population structure of Western Sandpipers across the breeding and non-breeding grounds. Using two different types of molecular markers, Amplified Fragment Length Polymorphisms (AFLPs; a type of genomic marker) and mitochondrial DNA sequence data, I examine the overall level of population genetic structure, as manifested in the degree of genetic divergence among sites, as well as possible broad-scale geographic patterns. Furthermore, I explore how the genetic structure relates to patterns of morphological variation (represented by body size) and implications for differential migration and life history strategies in this species.

Chapter 3:

Significant genetic population structure implies some level of isolation among breeding populations. But what can limit gene flow in a species as mobile as a migratory bird? To address this question, I examine another aspect of intraspecific variation by looking at geographic patterns in the characteristics of two different types of vocalizations: male songs and alarm calls. Bird song as a sexually selected signal may evolve faster than other traits and can play a crucial role in maintaining genetic differences or even in promoting divergence. As shorebird songs, unlike their oscine counterparts, do not appear to be learned, they have the potential to be closely associated with genetic differences. Alarm calls on the other hand, are not expected to be under sexual selection, and in fact tend to show similarities even across species (Miller 1996). They are therefore less likely to differ substantially among populations.

Chapter 4:

Following up on Chapter 3, Chapter 4 explores sources of intra- and inter- individual variation in the vocalizations of male Western Sandpipers. In particular, it takes a closer look at the relationships between vocalization frequencies and body size, but also considers other potential sources of vocal variation, including reproductive status and vocal individuality.

Chapter 5:

In this chapter, I further investigate male song by looking at its function in territorial defence. By testing male responses to song playbacks, this chapter also builds the basic pre-requisite for the playback experiments conducted in chapter 6.

Chapter 6:

In the last chapter, I examine the biological meaningfulness of geographic variation in male Western Sandpiper songs by studying whether males discriminate between songs from different breeding sites. I use male responses to song playbacks as a proxy to evaluate whether the degree of geographic variation in song is likely to be sufficiently large to potentially limit gene flow among breeding sites.

1.2. Synthesis of the main thesis results, their implications and suggestions for future research

1.2.1. Population structure in Western Sandpipers

To my knowledge, this is the first study that combines genetic, morphological and acoustic data to investigate the population structure of a migratory shorebird species. Geographical variation was evident in all of these and in conjunction, the results from the two different types of genetic markers, the morphological and the song data indicated low but significant population structure for Western Sandpipers. I found evidence for genetic population structure in both types of genetic markers, not only across the breeding, but also across the non-breeding grounds. Based on the AFLP data, isolation by distance mechanisms appear to play a defining role in creating the observed population structure in this species.

The mitochondrial control region sequences furthermore suggested that Western Sandpipers may have undergone a recent population expansion after experiencing a bottleneck. This appears to be a common pattern in arctic-breeding shorebirds and

presumably reflects range and demographic expansions after a period of confinement in a limited number of refugia during late Pleistocene glacial maxima (Wennerberg 2001; Buehler and Baker 2005; Pruett and Winker 2005; Rönkä et al. 2012; Lounsberry et al. 2013). In the case of the Western Sandpiper, one major haplotype dominated the star- shaped haplotype network, with remarkably few mutations separating many rare haplotypes. Expansion therefore likely took place out of a single refugium.

The short evolutionary time frames suggested by the low degree of sequence divergence in conjunction with the demographic expansion also imply that genetic population structure may be low because it is recent and insufficient time has passed for lineage sorting. Additionally, it is possible for the true level of isolation among populations to be obscured by population growth effects as these may counter-act genetic drift (Avise 2000). The results from this study contribute to the growing evidence that intraspecific divergence in many shorebird species is recent. Consequently, ancestral variation may be retained in genetic markers that do not experience strong selection pressures, which may explain why low or no population structure is seen even across species with diverging morphological or other traits (Rogers 1995; Avise 2000). The level of divergence observed in these markers, therefore, may not necessarily be indicative of the level of divergence in traits under selection nor accurately portray current levels of isolation among different populations in these species. This is not to say that all of these species are indeed undergoing divergence. It is quite possible that many of them are panmictic or nearly so and that trait variations primarily represent phenotypic plasticity. However, the fact that many of these species appear to have experienced recent population expansions suggests we should cautiously interpret low levels of population structure in markers commonly used in population genetics as they may not hold all the answers.

While the mitochondrial DNA sequence data was informative in terms of the demographic history of Western Sandpipers, and indicated that haplotype frequency differences occurred among sites, it did not reveal interpretable, large-scale geographic patterns. In contrast, the genetic structure revealed by the AFLP markers not only

appeared to reflect broad-scale geographic patterns but these patterns also showed correspondences with morphological variation. The patterns revealed by spatial Principal Component Analyses along with significant Mantel tests suggested that genetic divergence among Western Sandpiper populations may be primarily a consequence of isolation by distance processes. The primary structure across both the breeding and non- breeding grounds suggested a latitudinal pattern. These genetic patterns showed weak to strong correlations with body size on individual and population mean bases respectively. While no such correlation was detected among breeding females, and methodological concerns warrant a cautious interpretation of the morphological data, my results nevertheless suggest some level of congruence between genetic and morphological variation. Thus, differences in body size may be linked to genetic differences among populations. This relationship and the short evolutionary timeframes indicated by the mitochondrial DNA data, suggest body size in Western Sandpiper may be exposed to differential selection. Geographic variation in body size in this species may therefore represent adaptations to local conditions at different latitudes.

A pattern that appeared to reflect a latitudinal gradient also emerged from my analysis of song characteristics. Based on the results of the multivariate analysis, males at a northern breeding site appeared to sing songs that were higher in frequency and shorter in duration than those recorded at a southern breeding site, while songs from two centrally located sites displayed intermediate characteristics. These differences were quantitative, i.e. I did not observe clear differences in song structure across sites. Given the small geographic scale and presumed innate nature of these songs, the observed differences are nevertheless notable. Even though song and alarm call frequencies showed some correlation on an individual based level, alarm call characteristics did not appear to differ among breeding sites. Unlike alarm calls, songs are subject to sexual selection. As such they can be expected to be among the first traits to show geographical differences and at the same time may constitute key elements in promoting further differentiation (West-Eberhard 1983, Panhuis et al. 2001). The fact that geographical variation was evident in song characteristics further supports the hypothesis that traits

under selection may have begun to diverge in this species, irrespective of low overall levels of genetic population structure. The patterns I observed suggest that songs, unlike alarm calls, may be undergoing divergent selection, e.g. as a result of latitudinal differences in the strength of sexual selection. Alternatively, songs could be more strongly linked to a trait under selection. Body size immediately comes to mind, as song frequencies were significantly negatively correlated with a multivariate measure of body size. Alarm call frequencies merely showed a trend and therefore a possibly more tentative relationship with body size. While there is no immediately obvious, inherent expectation for song frequencies to be more strongly constrained by body size than alarm call frequencies, it is an intriguing possibility that deserves further exploration. On the other hand, alarm calls may simply be under stronger stabilizing selection that has thus far prevented divergence among sites.

As expected, given the postulated innate nature of shorebird songs, I found no evidence for different dialects, but found that geographic variation in song was congruent with genetic patterns. My results suggest that divergence in genetic markers and song characteristics may indeed be closely linked in shorebirds, making them a potentially interesting model for future research on the interplay between song and genetic divergence without the complicating factors of song learning. I demonstrate that the level of song divergence seen in shorebirds, though falling far short of the level of divergence commonly seen in songbird dialects, nevertheless has the potential to be of biological significance, by showing that male Western Sandpipers differentiate between songs recorded at the same site and those from a different breeding site. As I show in chapter 5, song also fulfills a territorial defence function indicating that it may be exposed to intra- in addition to inter-sexual selection. The effect of a song’s geographical origin, i.e. the elicited difference in male song response, was minimal, however. Given the small spatial scale involved in this study, this is not surprising and it seems fairly remarkable that even such a small effect can be observed. The sensitivity of shorebirds to even small-scale changes in vocalization characteristics could generally be high. In fact, while shorebird songs are clearly species-specific, presumably homologous (e.g. structural similarities)

elements may be evident even across species (Miller 1996), hinting at potentially fundamental differences in song evolution between shorebirds and songbirds. The degree to which female Western Sandpipers may make similar distinctions as males and in how far this may or may not influence their mate choices is, of course, unclear. Thus, further experiments are needed to gauge possible effects of this level of divergence in song on gene flow.

Overall, the observed low, but significant levels of genetic population structure in Western Sandpipers, in conjunction with geographic variation in song and morphological traits, may be an indicator for evolution under selective pressure rather than by genetic drift. I did not find evidence for clear-cut geographic barriers to gene flow, rather isolation by distance mechanisms appear to play a defining role in creating the observed population structure in this species. More pronounced genetic differentiation may thus be restricted to a few, key genes under divergent selection and selective pressures may be maintaining differences among sites despite ongoing gene flow. In arctic-breeding shorebirds, the influence of glacial maxima on population structure and dynamics may be particularly pronounced and play a decisive role in shaping the population structures of these species (e.g. Wenink et al. 1996; Buehler et al. 2006; Rönkä et al. 2012; Lounsberry et al. 2013; Trimbos et al. 2014). In accordance with this, the population structure and patterns of geographic variation seen in Western Sandpipers today are most likely the consequence of an evolutionarily recent population expansion that likely occurred in the wake of the last glacial maximum, in conjunction with subsequent adaptations to local conditions. In light of the gradual nature of the geographical variation and suggested gene flow, it appears unlikely that Western Sandpiper populations are undergoing a rapid divergence which might lead to more pronounced differences. This could change, however, if populations became more fragmented, e.g. due to a decrease in population size, which might interrupt gene flow.

1.2.2. Implications for migratory connectivity and differential migration

As the genetic markers generally did not show fixed differences (i.e. completely present in one while completely absent in another population) and the genetic population structure was low, it did not appear sensible to conduct assignment tests to directly investigate links between populations and evaluate the strength of migratory connectivity. My results therefore only allow indirect conclusions regarding the level of migratory connectivity and the likely origin of the observed patterns of differential migration with regard to body size in this species. Across the breeding and non-breeding grounds, the genetic differentiation based on AFLP markers showed corresponding latitudinal patterns. Furthermore, I found evidence that the latitudinal genetic variation is correlated with variation in body size in both cases. Overall, this suggests that individuals originating at more northerly located breeding sites also tend to be found further north during the non-breeding season, while southern breeders are found further south – effectively resembling a chain migration pattern but without clear-cut population boundaries.

One possible interpretation is that population genetic differences are instrumental in driving differential migration patterns on a large scale, by influencing migratory tendencies. This does not necessarily translate to strong migratory connectivity among individual breeding and non-breeding sites. Rather, individuals originating from breeding sites that are located towards the northern end of the breeding range may show migratory tendencies that increase their likelihood of spending the non-breeding season at sites, that, on average, tend to be located further north within the non-breeding range, while southern breeders may be more likely to migrate further south. Ultimately, such tendencies would allow for a substantial degree of mixing among populations, while still leading to the broader scale latitudinal patterns observed in this species. In fact, results from a stable isotope based study suggest such substantial mixing occurs, rendering strong migratory connectivity unlikely for this species (Franks et al. 2012). Ultimately, this suggests that genetic control of migratory tendencies is likely to be weak, allowing room for modification of migratory strategies via phenotypic plasticity.

If we look at the pattern from a different angle, however, it is conceivable for the population genetic differences to be the consequence of differential migration rather than being its driving force. It is possible that migratory behaviour itself is not directly under genetic control, but that the distance migrated is instead determined by an individual’s decision regarding which sites to remain at (i.e. when/where to terminate its migration). If site choices are mediated by an individual’s morphology, the observed latitudinal patterns could still emerge under certain conditions. For instance, smaller bodied individuals might choose to breed and to overwinter at more northerly located sites within the breeding and non-breeding ranges respectively, because conditions at these sites tend to be more favourable for smaller birds in both cases. The opposite might be true for larger individuals. If the morphological characteristics in question, e.g. body size, have a substantial genetic component, the offspring of these individuals, having inherited a similar body size, could show similar migratory tendencies. They would therefore be less likely to mate with larger individuals, which ultimately could create an isolation by distance based divergence pattern.

If and why smaller individuals might have a selective advantage at more northerly breeding sites and/or larger ones further south is unclear. The intensity of sexual selection might increase with latitude due to a decrease in breeding season length (Irwin 2000). Male advertisements involve aerial displays, thus small males could have a selective advantage at northern sites due to their increased manoeuvrability. This would not, however, provide an explanation for females under a site choice hypothesis. A number of hypotheses have been suggested to explain the morphological gradient on the non- breeding grounds, with food burial depth and predation danger hypotheses having received some empirical support (Nebel 2005; Nebel and Ydenberg 2005; Mathot et al. 2007; but see Franks 2012).

1.2.3. Conservation implications and future studies

While the Western Sandpiper is not currently an endangered species, they have been listed as a species of high conservation concern (Brown et al. 2001; Donaldson et al.

2001). Furthermore, there is a growing concern for shorebird species in general and many of the postulated challenges are likely to also affect Western Sandpipers. A major concern is habitat loss: coastal areas are subject to rapid developments and rising sea levels may not only affect non-breeding and stopover habitats, but may have significant impacts on breeding habitats in low-lying coastal areas (Brown et al. 2001; Donaldson et al. 2001; Shepherd et al. 2003; Terenzi et al. 2014). Relatively low overall levels of genetic population structure combined with a substantial amount of phenotypic plasticity with regard to migratory strategies may help to buffer Western Sandpiper populations against the loss of habitat in some areas which would be positive from a conservation perspective. If this is the case, a good conservation strategy might be to focus on important sites with large numbers of individuals. Nevertheless, it would be advisable to err on the side of caution. Some of the data from my study indicates some divergence may be taking place, particularly in traits under selection, and that there may be limits to phenotypic plasticity and thus flexibility in migratory behaviours on a larger scale. Therefore, an effort should be made to include sites located at different latitudes across the species’ breeding and non-breeding range to ensure the viability of different migratory and life history strategies. At present, no definitive conclusions can be drawn with regard to the level of divergence in genes that code for traits under selection and further studies are also needed to arrive at a more conclusive answer regarding migratory connectivity. Light-level geolocators and next-generation sequencing technologies could be utilized in future studies to obtain a more detailed understanding of Western Sandpiper migration strategies, which may in turn enable more finely tuned conservation strategies. A comprehensive approach that follows individual chicks from different populations into adulthood, combining ecological, physiological, genomic and light-logger data, while challenging, would have the potential to provide detailed insights that would allow us to gauge the relative importance of different contributing factors.

1.3. References

Avise, J. C. 2000. Phylogeography: the history and formation of species. Harvard University Press, Cambridge, MA.

Bart, J., S. Brown, B. Harrington, and R. I. G. Morrison. 2007. Survey trends of North American shorebirds: population declines or shifting distributions? Journal of Avian Biology 38:73-82.

Bell, C. P. 2005. Inter- and intrapopulation migration patterns: ideas, evidence and research priorities. Pages 41-52 in Birds of Two Worlds: The Ecology and Evolution of Migration. (R. Greenberg, and P. P. Marra, Eds.). The Johns Hopkins University Press, Baltimore.

Brown, R. G. B. 1962. The Aggressive and Distraction Behaviour of the Western Sandpiper Ereunetes mauri. Ibis 104:1-12.

Brown, S., C. Hickey, B. Harrington, and R. Gill, Eds. 2001. The U.S. Shorebird Conservation Plan, 2nd ed. Manomet Center for Conservation Sciences, Manomet, MA.

Buehler, D. M., and A. J. Baker. 2005. Population divergence times and historical demography in Red Knots and Dunlins. Condor 107:497-513.

Buehler, D. M., A. J. Baker, and T. Piersma. 2006. Reconstructing palaeoflyways of the late Pleistocene and early Holocene Red Knot Calidris canutus. Ardea 94:485-498.

Butler, R. W., G. W. Kaiser, and G. E. J. Smith. 1987. Migration Chronology, Length of Stay, Sex-Ratio, and Weight of Western Sandpipers, (Calidris mauri) on the South Coast of British Columbia. Journal of Field Ornithology 58:103-111.

Cristol, D. A., M. B. Baker, and C. Carbone. 1999. Differential migration revisited: latitudinal segregation by age and sex classes. Current Ornithology 15:33-88.

Donaldson, G. M., C. Hyslop, R. I. G. Morrison, H. L. Dickson, and I. Davidson, Eds. 2001. Canadian Shorebird Conservation Plan. Canadian Wildlife Service, Ottawa, Canada. .

Enríquez-Paredes, L. M., C. Vilanova, and G. Fernandez. 2012. Genetic diversity and population structure of wintering Western Sandpipers from the Sinaloa coast, Mexico. Journal of Field Ornithology 83:85-93.

Enríquez, L. M., and G. Fernandez. 2010. Genetic identity and population structure of nonbreeding Western Sandpiper in Mexico. Poster presented at the COS/AOU/SCO 2010 Joint Meeting, San Diego, CA.

Faaborg, J., R. T. 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. L. Merkord, E. Nol, S. I. Rothstein, T. W. Sherry, T. S. Sillett, F. R. Thompson, and N. Warnock. 2010. Conserving migratory land birds in the New World: Do we know enough? Ecological Applications 20:398-418.

Franks, S., D. B. Lank, and W. H. Wilson. 2014. Western Sandpiper (Calidris mauri), The Birds of North America Online. (A. Poole, Ed.). Cornell Lab of Ornithology, Ithaca. http://bna.birds.cornell.edu/bna.

Franks, S. E. 2012. Population connectivity and the causes and consequences of differential migration in a long-distance migratory shorebird, the western sandpiper. Ph.D. Thesis, Simon Fraser University, Burnaby.

Franks, S. E., D. R. Norris, T. K. Kyser, G. Fernandez, B. Schwarz, R. Carmona, M. A. Colwell, J. C. Sandoval, A. Dondua, H. R. Gates, B. Haase, D. J. Hodkinson, A. Jimenez, R. B. Lanctot, B. Ortego, B. K. Sandercock, F. Sanders, J. Y. Takekawa, N. Warnock, R. C. Ydenberg, and D. B. Lank. 2012. Range-wide patterns of migratory connectivity in the Western Sandpiper Calidris mauri. Journal of Avian Biology 43:155- 167.

Gonzalez-Voyer, A., and N. Kolm. 2011. Rates of phenotypic evolution of ecological characters and sexual traits during the Tanganyikan cichlid adaptive radiation. Journal of Evolutionary Biology 24:2378-2388.

Haig, S. M., C. L. GrattoTrevor, T. D. Mullins, and M. A. Colwell. 1997. Population identification of western hemisphere shorebirds throughout the annual cycle. Molecular Ecology 6:413-427.

Holmes, R. T. 1971. Density, Habitat and the Mating System of the Western Sandpiper (Calidris mauri). Oecologia 7:191-208.

Holmes, R. T. 1973. Social Behavior of Breeding Western Sandpipers Calidris mauri. Ibis 115:107-123.

Irwin, D. E. 2000. Song variation in an avian ring species. Evolution 54:998- 1010.

Lanctot, R. B., B. K. Sandercock, and B. Kempenaers. 2000. Do male breeding displays function to attract mates or defend territories? The explanatory role of mate and site fidelity. Waterbirds 23:155-164.

Lounsberry, Z. T., J. B. Almeida, T. Grace, R. B. Lanctot, J. Liebezeit, B. K. Sandercock, K. M. Strum, S. Zack, and S. M. Wisely. 2013. Range-Wide Conservation Genetics of Buff-Breasted Sandpipers (Tryngites subruficollis). Auk 130:429-439.

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.

Mathot, K. J., B. D. Smith, and R. W. Elner. 2007. Latitudinal clines in food distribution correlate with differential migration in the Western Sandpiper. Ecology 88:781-791.

Miller, E. H. 1996. Acoustic Differentiation and Speciation in Shorebirds. in Ecology and evolution of acoustic communication in birds (D. E. Kroodsma, Miller, E. H., Ed.). Comstock/Cornell University Press, Ithaca, New York.

Morrison, R. I. G., Y. Aubry, R. W. Butler, G. W. Beyersbergen, G. M. Donaldson, C. L. Gratto Trevor, P. W. Hicklin, V. H. Johnston, and R. K. Ross. 2001. Declines in North American shorebird populations. Wader Study Group Bulletin 94:34- 38.

Morrison, R. I. G., C. Downes, and B. Collins. 1994. Population Trends of Shorebirds on Fall Migration in Eastern Canada 1974-1991. Wilson Bulletin 106:431- 447.

Nebel, S. 2005. Latitudinal clines in bill length and sex ratio in a migratory shorebird: a case of resource partitioning? Acta Oecologica-International Journal of Ecology 28:33-38.

Nebel, S., D. B. Lank, P. D. O'Hara, G. Fernandez, B. Haase, F. Delgado, F. A. Estela, L. J. E. Ogden, B. Harrington, B. E. Kus, J. E. Lyons, F. Mercier, B. Ortego, J. Y. Takekawa, N. Warnock, and S. E. Warnock. 2002. Western sandpipers (Calidris mauri) during the nonbreeding season: Spatial segregation on a hemispheric scale. Auk 119:922- 928.

Nebel, S., and R. Ydenberg. 2005. Differential predator escape performance contributes to a latitudinal sex ratio cline in a migratory shorebird. Behavioral Ecology and Sociobiology 59:44-50.

O'Hara, P. D., G. Fernandez, F. Becerril, H. de la Cueva, and D. B. Lank. 2005. Life history varies with migratory distance in Western Sandpipers Calidris mauri. Journal of Avian Biology 36:191-202.

O'Hara, P. D., G. Fernandez, B. Haase, H. de la Cueva, and D. B. Lank. 2006. Differential migration in Western Sandpipers with respect to body size and wing length. Condor 108:225-232.

Panhuis, T. M., R. Butlin, M. Zuk, and T. Tregenza. 2001. Sexual selection and speciation. Trends in Ecology & Evolution 16:364-371.

Price, T. 1998. Sexual selection and natural selection in bird speciation. Philosophical Transactions of the Royal Society of London Series B-Biological Sciences 353:251-260.

Price, T. 2007. Speciation in Birds. Roberts and Company Publishers, Greenwood Village, Colorado.

Pruett, C. L., and K. Winker. 2005. Biological impacts of climatic change on a Beringian endemic: Cryptic refugia in the establishment and differentiation of the rock sandpiper (Calidris ptilocnemis). Climatic Change 68:219-240.

Rogers, A. R. 1995. Genetic Evidence for a Pleistocene Population Explosion. Evolution 49:608-615.

Rönkä, N., L. Kvist, V. M. Pakanen, A. Rönkä, V. Degtyaryev, P. Tomkovich, D. Tracy, and K. Koivula. 2012. Phylogeography of the Temminck's Stint (Calidris temminckii): historical vicariance but little present genetic structure in a regionally endangered Palearctic wader. Diversity and Distributions 18:704-716.

Sandercock, B. K. 1997. Factors affecting the breeding demography of Western Sandpipers (Calidris mauri) and Semipalmated Sandpipers (C. pusilla) at Nome, Alaska. Ph.D. Thesis, Simon Fraser University, Burnaby.

Seddon, N., C. A. Botero, J. A. Tobias, P. O. Dunn, H. E. A. MacGregor, D. R. Rubenstein, J. A. C. Uy, J. T. Weir, L. A. Whittingham, and R. J. Safran. 2013. Sexual

selection accelerates signal evolution during speciation in birds. Proceedings of the Royal Society B: Biological Sciences 280.

Shepherd, P. C. F., L. J. Evans Ogden, and D. B. Lank. 2003. Integrating marine and terrestrial habitats in shorebird conservation planning. Wader Study Group Bull. 100 40-42.

Stein, R. W., G. Fernandez, H. de la Cueva, and R. W. Elner. 2008. Disproportionate bill length dimorphism and niche differentiation in wintering western sandpipers (Calidris mauri). Canadian Journal of Zoology-Revue Canadienne De Zoologie 86:601-609.

Terenzi, J., M. T. Jorgenson, and C. R. Ely. 2014. Storm-Surge Flooding on the Yukon-Kuskokwim Delta, Alaska. Arctic 67:360-374.

Trimbos, K. B., C. Doorenweerd, K. Kraaijeveld, C. J. M. Musters, N. M. Groen, P. de Knijff, T. Piersma, and G. R. de Snoo. 2014. Patterns in Nuclear and Mitochondrial DNA Reveal Historical and Recent Isolation in the Black-Tailed Godwit (Limosa limosa). PLoS ONE 9:12.

Webster, M. S., and P. P. Marra. 2005. The Importance of Understanding Migratory Connectivity and Seasonal Interactions. in Birds of Two Worlds: The ecology and Evolution of Migration (R. Greenberg, and P. P. Marra, Eds.). The Johns Hopkins University Press, Baltimore.

Webster, M. S., P. P. Marra, S. M. Haig, S. Bensch, and R. T. Holmes. 2002. Links between worlds: unraveling migratory connectivity. Trends in Ecology & Evolution 17:76-83.

Wenink, P. W., A. J. Baker, H. U. Rosner, and M. G. J. Tilanus. 1996. Global mitochondrial DNA phylogeography of holarctic breeding dunlins (Calidris alpina). Evolution 50:318-330.

Wennerberg, L. 2001. Genetic Variation and Migration of Waders. Ph.D., Lund University, Sweden, Lund.

West-Eberhard, M. J. 1983. Sexual selection, Social Competition, and Speciation. Quarterly Review of Biology 58:155-183.

Ydenberg, R. C., R. W. Butler, D. B. Lank, B. D. Smith, and J. Ireland. 2004. Western sandpipers have altered migration tactics as peregrine falcon populations have recovered. Proceedings of the Royal Society of London Series B-Biological Sciences 271:1263-1269.

Chapter 2. Population Genetics of Western Sandpipers

2.1. Abstract

Seasonal migration has the potential to either promote or hinder intraspecific divergence. Many arctic-breeding shorebirds are long-distance migrants that primarily rely on coastal habitats during the non-breeding season. The patchy, but reliable nature of these non-breeding habitats could lead to strong migratory connectivity and local adaptation, potentially facilitating divergence among populations. Western Sandpipers are long-distance migrants that show morphological and life history differences across their non-breeding range. We investigated their genetic population structure, its relationship with body size and consequent implications for the observed differential migration patterns. The sequencing of a 681bp long fragment of the mitochondrial control region revealed one common and many closely related, rarer haplotypes. The resulting star-shaped median-joining haplotype network, significant negative Tajima’s D

(-2.34) and Fu’s FS (-28.10) values (P<0.001), and the low R2 value (0.02, P<0.01) suggested a recent demographic expansion. Similarly to other shorebird species, this may

have taken place after the last glacial maximum. Haplotype frequency based FST values were significant (FST=0.2 (breeding) and 0.21 (non-breeding), P<0.001), while molecular

distance based ΦST values were not. Amplified Fragment Length Polymorphisms (AFLP)

revealed low, but significant population structure (band frequency based FST=0.027 (breeding) and 0.016 (non-breeding), P<0.001 and P<0.05) and suggested an isolation by distance based, latitudinal divergence pattern among breeding and non-breeding sites. While we found no evidence for fixed genetic differences among juveniles likely differing in life-history strategies or in migratory direction, our data did suggest a

moderate level of congruence between genetic and body size based geographic patterns. Genetic and morphological principal component scores were correlated across breeding and non-breeding grounds (except for breeding females). Overall, our results suggest that genetic population structure, morphology and latitudinal segregation are likely interrelated and could be explained by a gradual ’chain’-like migration pattern with weak migratory connectivity. Future studies are required to clarify causal relationships.

2.2. Introduction

A migratory life-style is likely to affect a species’ biology in manifold ways. When it comes to patterns of genetic divergence within a species, the effects of migration are double-edged as it theoretically has the potential to promote divergence, but could also constitute a hindrance (Irwin and Irwin 2005; Webster and Marra 2005; Winker 2010). A genetically controlled migratory program could have diversifying effects via relatively fixed migratory routes that result in increased natal philopatry and winter site fidelity (Webster and Marra 2005; Henningsson and Alerstam 2006). Greater philopatry may restrict gene flow among different breeding sites while both philopatry and non- breeding site fidelity conceivably favour the evolution of adaptations to conditions at the respective sites (Webster et al. 2002; Henningsson and Alerstam 2006). As a consequence, traits – including those under sexual selection – may diverge among populations using different sites, promoting reproductive isolation (Irwin and Irwin 2005). Other migration-related factors, such as differences in arrival timing, pair- formation at non-breeding sites or inferior migratory behaviour in hybrids may also reduce gene flow among populations or subspecies (Bensch 1999; Esler 2000; Irwin and Irwin 2005; Bensch et al. 2009; Irwin 2009). Conversely, a migratory species’ inherent mobility may have the opposite effect as it has the potential to reduce the negative effects that physical barriers or distances may have on gene flow among breeding populations (Winker 2010; Haig et al. 2011; Liu et al. 2012). In general, strong migratory connectivity among breeding and non-breeding sites is likely to be conducive to diversification, whereas weak or no connectivity and the resulting mixing of populations

would be expected to impede the evolution of local adaptations, rendering divergence less likely.

Many arctic-breeding shorebirds are long-distance migrants and morphological differences among populations are not uncommon (e.g. (Engelmoer and Roselaar 1998; Wennerberg et al. 2002; Gratto-Trevor et al. 2012). For those species that winter coastally, the patchy distribution of their coastal wintering sites may favour patterns of relatively strong migratory connectivity, linking breeding and non-breeding populations (Kraaijeveld 2008). Consequently, differences in morphological features such as bill length may evolve among disparate populations in response to differing ecological conditions among sites. However, morphological differences do not appear to show a close correspondence with genetic differences in all species. In fact, some species show little or no evidence of genetic differentiation despite marked morphological variation among individuals at different sites (e.g. Wennerberg et al. 2002; Buehler and Baker 2005; Miller et al. 2013).

One shorebird species for which morphological differences across non-breeding sites are suggestive of local adaptation, but for which stable isotope data indicates weak connectivity between breeding and non-breeding sites, is the Western Sandpiper (Calidris mauri; Nebel 2005; O'Hara et al. 2006; Stein et al. 2008; Franks et al. 2012). This small shorebird is a long-distance migrant that breeds in Western Alaska and easternmost Siberia and winters along the Pacific and Atlantic coasts from central or southern North America to northern South America (Franks et al. 2014). On the breeding grounds, previous studies indicated high adult return rates but lower natal philopatry (ca. 3% return rate) on small spatial scales (i.e. study plots ranging from a few hectares to 4km2 in size; Holmes 1971, Sandercock 1997). Across their non-breeding range, Western Sandpipers show not only morphological but also life history differences: individuals wintering further north tend to have shorter wings and culmen (and/or smaller body sizes) than those wintering further south and juveniles that spend their first winter at northern sites appear to be more likely to return to the breeding grounds in their first year of life than

those wintering further south (Nebel 2005; O'Hara et al. 2005; O'Hara et al. 2006; Stein et al. 2008; Franks 2012).

If these observed patterns arise due to genetically different populations, we would expect to see comparable differences in morphology across the breeding range as predicted by Elner and Seaman (2003), along with a corresponding genetic population structure. A detailed study on Western Sandpiper breeding ground morphology is currently lacking. However, measurements taken during spring migration at stopover sites close to breeding grounds and at breeding sites along the Chuckchi Sea coast indicated the possibility of variability in morphological traits across populations, but were not consistent across years (Senner et al. 1981). Even less is known about the genetic population structure of this species and what little information is available is almost exclusively restricted to a few sites on the non-breeding grounds. The respective studies suggest some degree of genetic differentiation may be present across non-breeding populations at different sites and/or in different habitats, but are too limited in scope to allow conclusions to be drawn regarding the origin of the aforementioned morphological variation (Haig et al. 1997; Enríquez and Fernandez 2010; Enríquez-Paredes et al. 2012). A correspondence among morphological, genetic and, possibly, life history patterns across the non-breeding grounds would lend support to a genetic adaptation hypothesis – particularly, if corresponding patterns could also be found among breeding ground sites. However, the lack of evidence for strong migratory connectivity in this species (Franks et al. 2012), calls into question how such a correlation could have arisen with regard to the non-breeding grounds. It appears unlikely that different breeding populations could have evolved genetic adaptations to conditions at specific non-breeding sites given that individuals from different breeding sites appear to mix to a large degree on the non- breeding grounds. Nevertheless, population genetic differences, particularly if accompanied by morphological divergence, might directly or indirectly influence an individual’s migratory behaviour. Genetically and morphologically different individuals might show slight differences in their non-breeding site preferences or employ migratory strategies that fit their particular morphology. Such preferences or strategies might be

based on behavioural decisions or physical abilities, in which case a genetic influence might be mediated indirectly through the morphological differences. Alternatively, they might be more directly influenced by a genetically driven, endogenous migratory program (Pulido 2007; Liedvogel et al. 2011) or result from a combination of direct and indirect effects.

A lack of genetic and morphological divergence on the breeding grounds, on the other hand, would indicate that the morphological and life history patterns across the non- breeding range can not be attributed to or explained by population genetic differences. In this case, other mechanisms, such as phenotypic plasticity, might be driving the observed differential migration in this species. For instance, the observed north-south gradient on the non-breeding grounds could be driven primarily by behavioral decisions with individuals distributing themselves according to their individual morphological characteristics and physical condition. Under this alternative hypothesis, environmental effects might be primarily responsible for the variability in morphological traits: an individual’s morphological and physiological characteristics might be shaped by conditions experienced e.g. during development which would ultimately determine its migratory and life history strategy (Franks et al. 2012). Correspondingly, we would not expect to find genetic patterns that matched the apparent morphological cline across the non-breeding range.

In addition to the latitudinal pattern, differential migration may also occur between West and East Coast sites. Two of the above-mentioned studies indicate possible genetic differences between individuals migrating to Atlantic Coast versus those migrating to Pacific Coast non-breeding sites (Haig et al. 1997; Enríquez and Fernandez 2010). Such differences are not unexpected given the likely differences in migratory routes and direction involved for these populations, and highlight additional possible patterns of divergence and connectivity for this species.

Whether shorebird migration patterns and migratory strategies are primarily driven by population genetic origins or are largely shaped by environmental conditions

and behavioural plasticity, has important implications for the development of effective conservation strategies (Haig et al. 1997; Lounsberry et al. 2013). Yet, for most species little is known about the extent to which population genetic structures influence migration. Given the observed latitudinal and possible East-West Coast patterns in Western Sandpipers, we examined potential population genetic correlates of differential migration in this species. As population genetic differences ultimately originate on the breeding grounds, we investigated the genetic population structure across the breeding grounds in western Alaska and easternmost Siberia. Additionally, we tested whether there was evidence for significant genetic differences among groups of individuals at different non-breeding sites, and if so, in how far these differences were concordant with expected patterns based on morphological clines, life history patterns and likely differences in migratory directions. Lastly, we checked for correspondences between genetic profiles and morphological traits. As illustrated above, the following patterns would suggest that the morphological clines on the non-breeding grounds are at least partially driven by population genetic differences: significant genetic differentiation on the breeding grounds, corresponding genetic differences among individuals at different non-breeding sites, and a correspondence between genetic and morphological patterns. A lack of differentiation, on the other hand, would suggest that other mechanisms are likely responsible in driving the observed morphological patterns. Similarly, a clear and consistent genetic divergence among individuals with different life history strategies or migratory directions would suggest a direct influence of an individual’s genetic background on these behaviours.

2.3. Methods

2.3.1. Sampling and DNA extraction

We analyzed blood samples that were collected at one breeding site in eastern Russia, on the Chukotka Peninsula (67°03'N, 174°35'W; 2009) and at four breeding sites in western Alaska, near Barrow (71°16'N, 156°35'W; 2008-2010), Kotzebue (66°50'N,

162°34'W; 2009), Nome (64°26'N, 164°56'W; 2008-2009), and in the Yukon- Kuskokwim Delta (YK-Delta; 61°21'N, 165°07'W; 2000, 2006, 2009; Fig. 2-1). Individuals were caught using walk-in traps and blood samples obtained from their brachial veins. We did not differentiate among adult (>1 year of age) and first-year breeders (≈1 year).

We also analyzed blood samples from up to eight non-breeding sites where individuals were caught using mistnets during the 2008/2009 non-breeding season. These included sites sampled along the Pacific Coast: in California (Humboldt Bay 40°50'N, 124°05'W, San Francisco Bay 38°06'N, 122°24'W), Mexico (Bahia Santa Maria, 24°56'N, 107°54'W; La Paz 24°06'N, 110°22'W), Panama (Costa del Este, 9°00'N 79°27'W) and Ecuador (Salinas, 2°15'S, 80°56'W and 2°06'S, 80°44'W, treated as one site), as well as two sites on the Atlantic Coast in Eastern Mexico (Ría Lagartos, 21°36'N, 87°59'W) and Puerto Rico (Cabo Rojo National Wildlife Refuge, 17°58'N, 67°12'W). The samples from the two Californian sites were pooled for analyses.

Individuals were sexed on the basis of their culmen length, following Page and Fearis (1971), or genetically (particularly where culmen length fell in or close to the small region of overlap in culmen length between the sexes). For breeding individuals the sex of the breeding partner was also used to determine sex in some cases. Non-breeding individuals were aged as adult (>1 year of age) or juvenile (<1 year of age) on the basis of their body plumage, inner median covert edge color and flight feather wear (O'Hara 2003; Franks 2012). Samples from the Russian site were stored in ethanol; all other samples were stored in ‘Queen’s Lysis buffer’ (0.01 M Tris, 0.01M NaCl, 0.01 M EDTA, 1% n-lauroylsarcosine, pH 7.5; Seutin et al. 1991), or ‘Longmire’ buffer (Longmire et al. 1997). A standard phenol-chloroform protocol was used to extract DNA which was then re-suspended in TE-Buffer (10 mM Tris-HCl, 1 mM EDTA, pH 8.0). Final sample sizes for each site and type of analysis are given in Table 2-1.

2.3.2. mtDNA sequencing

We employed the following primers, developed by Enríquez-Paredes et al. (2012) for Western Sandpipers, to amplify a 681bp long fragment of the central domain of the mtDNA control region: CmCRFa (5´-CCCCCATACTACATACCATC-3´) and CmCRR (5´GTCCCACAAGCATTCATT-3´). Polymerase chain reactions (PCR) were conducted in 10μL volumes which contained 1x PCR buffer, 1.5mM MgCl2, 0.2mM dNTP mix, 0.5μM of each primer, and 0.4 U of Taq DNA Polymerase. Reactions were run under the following thermocycling profile: 94°C (3min), 35 cycles of 94°C (30s), 53.4°C (30s) and 72°C (30s), followed by 72°C (10min) and a cool-down to 4°C. Samples were sent to the McGill University and Génome Québec Innovation Centre (Montréal, Canada) for purification and sequencing. Of the 135 sequenced samples, 15 samples did not yield a full length sequence and were excluded from further analyses.

2.3.3. AFLP genotyping

Our AFLP procedure was based on a modified version of the Vos et al. (1995) protocol as described in Toews and Irwin (2008). The extracted DNA underwent digestion with restriction enzymes (EcoRI and MseI) prior to the ligation of E- and M- adaptors to the restriction fragments. A pre-amplification round in which a PCR was run with primers that contained one additional, arbitrary base-pair (T-C), but were otherwise complementary to the adaptors, was followed by selective amplification runs. In these selective runs, we used different combinations of M- and fluorescently marked E-primers with 3 additional, selective nucleotides (4 in the case of one M-primer) in a ‘touch-down’ PCR (see Table 2-5 for a list of the respective primer combinations). The generated PCR products were run in 6.5% polyacrylamide gels on a LI-CORE-4300 DNA analyzer (LI- COR, Lincoln, Nebraska). We scored the resulting AFLP band profiles as presence/absence data manually in the program SAGA (version 2.0, LI-COR, Lincoln, Nebraska). Only those bands that were variable and were both present and absent in two or more individuals were included. To check for consistency and repeatability of the scored markers, 23% of the samples were re-extracted, re-run and re-scored for each

marker. We retained only markers with high repeatability (>95%) for analysis. Similarly, any samples with inconsistent and highly deviant band patterns or high failure rates were excluded from the analysis.

In addition to the analysis described above in which each sample was run individually, we also employed a pooled approach, in which individuals from the same non-breeding site were combined. This approach was used to more efficiently scan a large number of primer combinations in a targeted search for markers that might correspond to fixed differences between juveniles with different life history and migration strategies. Due to sample size limitations we restricted our analysis to juvenile females. We chose samples from three sampling areas to represent juveniles wintering on the northern Pacific Coast (California, n=5), the southern Pacific Coast (Panama, n=8) and the central Atlantic Coast (Yucatan, n=8). For each of these three regions, the respective samples were pooled into one PCR reaction and subsequently run in one lane per region on the polyacrylamide gel (following a method developed by A. Brelsford (pers. comm.)). We tested 82 primer combinations (Appendix Table 2-9), which yielded several thousand potential markers. As a follow-up, any primer combination that yielded bands absent in at least one regional lane and with a clear and strong presence in at least one of the other lanes, was re-run normally with each sample in separate PCR reactions and lanes to confirm or refute a possible fixed difference.

2.3.4. Morphometric data collection:

An individual’s exposed culmen (from the tip to the edge of feathering) was measured to the nearest 0.1 mm using calipers. The straightened and flattened maximum wing chord (hereafter referred to as ‘wing’) was measured with a ruler to the nearest millimeter. Relaxed wing measurement data apparently taken in the YK-Delta in 2000 were converted using a conversion equation based on data from 2001 when both types of measures were taken. Individuals with missing wingtips or molting outermost primaries were excluded from morphological analyses. Tarsometatarsus (hereafter: ‘tarsus’) was measured with regular calipers or customized tarsus calipers to the nearest 0.1 mm and

was measured either as ‘diagonal’ (including the lower joint and measuring to the indentation below the upper joint) or as ‘full’ (including both joints) tarsus measurement, or both. In order to adjust ‘diagonal’ tarsus measurements to represent ‘full’ measurements, we calculated a conversion equation based on the measurements of individuals for which both types of measures were available.

2.3.5. Data analysis

mtDNA:

Sequences were aligned using the BioEdit software (version 7; Hall 1999) and checked manually. We used the program Arlequin (version 3.5; Excoffier and Lischer 2010) to calculate standard diversity indices (number of haplotypes, haplotype and nucleotide diversities). We investigated demographic processes by calculating Tajima’s D and Fu’s Fs in Arlequin and Ramos-Onsins and Rozas R2 statistic in DnaSP (Librado and Rozas 2009). To visualize genetic distances among haplotypes, a median-joining haplotype network was created in PopART (version 1.7; http://popart.otago.ac.nz). To investigate population genetic structure we conducted AMOVAs in Arlequin, with males and females considered separately for non-breeding sites only. Analogous to Enríquez-

Paredes et al. (2012), we employed both a frequency (FST) and a distance based (ΦST) approach.

A high proportion of rare haplotypes can increase the level of uncertainty when estimating the degree of population differentiation, particularly when sample sizes are small (Formia et al. 2007). To address this issue, we repeated the AMOVAs on a subset of the data. For this subset we only included those haplotypes which were present in two or more individuals in our dataset or corresponded to haplotypes found in the Enríquez- Paredes et al.(2012) study (two instances) and therefore were unlikely to be extremely rare. To account for false discovery rates (FDRs) relating to multiple comparisons, we applied q-values to population pairwise comparisons (Storey 2002). All pairwise FST values with p, q and local FDR values below 0.01 were considered significant; applying

the stricter cut-off further reduced the likelihood of spurious results due to rare alleles in conjunction with moderately low sample sizes.

AFLP:

We conducted AMOVAs, Mantel tests and spatial Principal Component Analyses on the breeding and non-breeding ground data. An initial AMOVA indicated no significant genetic differences between juvenile and adult individuals of the same sex that were caught at the same wintering site. We therefore pooled the two age classes for statistical analyses except where indicated otherwise.

As AFLPs are dominant markers, heterozygotes cannot (easily) be distinguished from individuals that are homozygous for band presence. Hence, allele frequencies have to be estimated to calculate measures of genetic differentiation that are allele-frequency based. Alternatively, band-frequency based distance measures can be obtained. The latter calculate distances on the phenotypic basis of observed presences/absences and therefore do not require allele frequencies to be estimated, but these are not directly comparable to values obtained with e.g. codominant markers. To facilitate comparability with other

studies, we calculated overall FST values (or their analogues) using three different methodological approaches: two allele-based and one band-based estimate. In the first approach we used the program TFPGA version 1.3 (Miller 1997) to calculate Weir and

Cockerham’s (1984) FST analogue θ. In this case, allele frequencies were estimated with Lynch and Milligan’s (1994) Taylor expansion estimate. This estimate has been shown to perform well in simulations (Bonin et al. 2007), but assumes Hardy-Weinberg- Equilibrium (HWE) and the presence of only two alleles for each marker. Secondly, we employed the Bayesian approach implemented in Hickory’s (version 1.1) f-free model, which makes no assumptions about the level of inbreeding (Holsinger et al. 2002).

Finally, we obtained band-based FST measures through Arlequin’s AMOVA procedure which was run on a band-based dissimilarity matrix. We used a Dice/Sørensen dissimilarity coefficient calculated in FAMD version 1.3 (Schlüter and Harris 2006). Dice/Sørensen or the related Jaccard coefficient are considered more appropriate

measures of (dis)similarity than simple-matching coefficients for AFLPs, as they ignore shared absences which are more likely to be subject to homoplasy (Bonin et al. 2007;

Gaudeul et al. 2012). The last method was also used to obtain pairwise FST estimates among sampling sites. As above, we accounted for multiple comparisons by applying q

values and considered all pairwise FST values with p, q and local FDR values below 0.05 significant.

To specifically test for possible clinal or isolation by distance patterns in genetic differentiation, we conducted reduced major axis regressions and Mantel randomization tests as implemented in the Isolation By Distance Web Service using all AFLP markers (Jensen et al. 2005). Given our geographically clustered sampling design, the analysis was conducted on a population pairwise genetic distance matrix. To avoid issues with non-independence we pooled all samples collected at the same site, regardless of sex. We utilized the Cavalli-Sforza and Edwards chord distance (Takezaki & Nei formula; Takezaki and Nei 1996) – a geometric distance which does not rely on specific evolutionary model assumptions – which we calculated using FAMD and the ‘Bayesian non-uniform prior’ setting for allele frequency estimation. Geographic distances between sampling sites were obtained from GPS points as great-circle distances.

AMOVA and associated FST values are useful tools to detect and quantify the level of genetic structure but do not directly provide readily interpretable information on large-scale geographic patterns. Mantel tests explore only one aspect of spatial structure (distance). To further explore geographical variation in genetic population structure and obtain more detailed information on spatial patterns, we performed a spatial Principal Component Analysis (sPCA), a technique that has been used in a number of population genetics studies spanning a wide range of organisms (e.g. Barr et al. 2015; Rouger and Jump 2015; Vander Wal et al. 2013). One of the advantages of ordination techniques, such as Principal Component Analysis, is that they do not require specific population model assumptions (e.g. Hardy-Weinberg equilibrium), nor do they force individuals into discrete groups, a procedure that may not be appropriate in the case of clinal variation (Jombart et al. 2008; Shirk 2009). We chose to conduct a sPCA, because this type of

multivariate analysis explicitly takes spatial information (in the form of spatial proximities of individuals) into account. In this approach distance does not enter as a direct variable but is used to define a connection network (i.e. networks that represent spatial relationships) which is then used to compute the spatial autocorrelation of allelic frequencies (via Moran’s I). Essentially, it optimizes the product of spatial autocorrelation and variance (Jombart et al. 2008). As such it outperforms PCA (which only maximizes variance irrespective of spatial information) in the representation of spatial genetic patterns and hence should be preferred when spatial patterns are of interest (Jombart 2008; Jombart et al. 2009; Teich et al. 2014). The sPCA method allows the detection of both local (negative spatial autocorrelation) and global (positive spatial autocorrelation) patterns of spatial structure in the genetic data – represented by negative and positive eigenvalues respectively, and their discernment from random noise. Genetic differences between spatially distinct groups, intermediates or geographical clines are represented by the global structure while notable genetic divergence among spatially close individuals emerges as a local pattern.

We conducted the sPCA using the R package adegenet (Jombart 2008, Jombart et al. 2008). We analyzed breeding and non-breeding data separately and, as our sampling design was not evenly distributed, used nearest-neighbour networks to reflect spatial relationships. GPS points were translated into UTM coordinates and jittered by 100m (i.e. a small amount of noise was added) to generate unique sampling locations for all individuals. We used the Monte Carlo based global and local tests provided in the R adegenet package to test whether at least one significant global or local pattern that should be interpreted is present in our data. These significance tests are based on the assumption that in the case of a true spatial pattern a large number of alleles should show correlations with spatial vectors (Jombart et al. 2008).

Correspondence of Morphology and AFLP data:

To obtain an overall measure of size and/or shape, we conducted a Principal Component Analysis using the three morphological measurements (wing, culmen and tarsus length). We then investigated whether the resulting morphological Principal Component scores correlated with the AFLP sPC1 scores obtained in the spatial PCA for the breeding and non-breeding grounds respectively. Separate analyses were performed for males and females, using individuals for which both types of data were available.

We also conducted a joint analysis of breeding and non-breeding samples. Due to its spatial nature, a sPCA on the genetic data could not be conducted directly using both sample types. We therefore performed a regular PCA using those AFLP markers that loaded strongly onto the non-breeding sPC1. The loading value cut-off ( +/- 0.13) was chosen based on a relative drop across loading values that still allowed retention of a substantial portion of the markers (>25%), and resulted in the inclusion of 22 (out of 80 non-breeding markers). To obtain a more representative measure of body size for each site, we conducted the morphological PCA on all individuals with complete morphometric data, including those for which genetic data was not available. We therefore examined the relationship of genetic and morphological data based on breeding and non-breeding site means. Means were calculated separately for each sex and entered into a correlation analysis. For this analysis only, individuals were sexed solely based on their culmen length and individuals whose culmen length measured between 24.3 and 24.7mm (inclusive) were excluded from the analysis (3.7%). To more clearly depict the relationship between latitude and the genetic scores from the joint PCA conducted for the morphology correlation analysis, we also plotted mean AFLP PC1 scores (both sexes) against latitude for both breeding and non-breeding sites.

2.4. Results

Sample sizes for all data types (mtDNA, AFLP and morphometric) are given in Table 2-1.

2.4.1. mtDNA

The sequencing of the 681bp long section of the mtDNA control region yielded 36 variable sites, resulting in 40 different haplotypes among the 120 samples that were successfully sequenced (Appendix Table S2-7). Substitutions consisted primarily of transitions with only a few transversions (33 vs. 4). The predominant haplotype (Hap01) was present in 45.8 % of individuals and corresponded to the most common haplotype found in a study on non-breeding Western Sandpipers at a site in western Mexico (Enríquez-Paredes et al. 2012). Of the remaining haplotypes, 24 were only present in single individuals in our dataset (singletons; Table 2-2) and 15 were present in two or more individuals. Two of the former and 6 of the latter were also found in the Mexican study along with the most common haplotype. In sum, only 9 of the found haplotypes were identical between the two studies, whereas 17 of the polymorphic sites corresponded between them. The resulting higher number of haplotypes (72) than polymorphic sites (48) thus suggests homoplasy. The overall haplotype diversity was moderately high (H= 0.786 +/- 0.040 [S.D.]), reflecting the relatively high number of haplotypes, while nucleotide diversity was low (π= 0.0023 +/- 0.0015 [S.D.]), indicating low levels of sequence divergence. The suggested close relationship between haplotypes is further demonstrated by the median-joining haplotype network. The network exhibited a star-shaped pattern with few substitutions (1-4) separating rarer haplotypes from the central, most prevalent one (Fig. 2-2). The star shape suggests a recent population expansion, a demographic scenario that is further supported by the negative Tajima’s D (-

2.34) and Fu’s FS (-28.10) values, both of which were highly significant (P < 0.001), and the low R2 value (0.02, P < 0.01).

Consistent with the apparent lack of phylogenetic structure suggested by the haplotype network, no population structure could be detected among the different breeding sites using a molecular distance-based AMOVA approach, nor was such structure evident among non-breeding sites (Table 2-3). The frequency-based AMOVA, however, indicated highly significant differences in haplotype-distribution among the different breeding sites, attributing 20% of the variation in haplotype frequencies to the

among site component (Table 2-3). Among non-breeding sites we found a nearly equivalent FST value of 0.21. Pairwise FST values ranged from 0.09 to 0.31 and most comparisons not involving the two populations with the smallest sample sizes (California females, Puerto Rico males) were significant (Table 2-4). The removal of all singleton

haplotypes from the analysis did not lead to a decrease in overall FST values (Table 2-3). Furthermore, solely one pairwise comparison became non-significant and another significant after singleton removal.

2.4.2. AFLP:

We scored 106 and 80 markers for breeding and non-breeding sites respectively, choosing only clear bands with a high repeatability (>95%) for analysis. One band was found exclusively at the Barrow breeding site, but occurred in only two individuals. None of the markers exhibited a fixed difference among breeding sites. Among the non- breeding samples, two of the markers were present in all samples of one sex at one site while being completely absent in the opposite sex at another site. However, in both cases sample sizes were low (California females (n=4) vs. Yucatan males (n=6) and Panama males (n=4)) and at each of the sites at which the respective marker was absent in one sex it did occur in the other sex, making it unlikely that this represents a genuine fixed difference among populations. Furthermore, the run of 82 primer combinations – yielding thousands of potential markers – in the pooled lane approach identified no fixed differences among juveniles with a postulated difference in life-history strategy or migratory direction.

Both band and allele frequency based estimates of population structure indicated

low, but significant differentiation among breeding sites with overall FST values ranging

from 0.027 to 0.052 (Table 2-3). Among non-breeding sites, a slightly lower FST value was suggested by both band and Bayesian allele frequency based estimates, while the non-Bayesian allele frequency based estimate indicated a higher value of 0.145 (Table 2- 3). Pairwise, among-site comparisons that were based on band frequencies showed significant differences both within and among the breeding and non-breeding range with

pairwise FST values ranging from slightly negative values, indicating no differentiation, to 0.1 (Table 2-6).

Mantel tests revealed a significant positive correlation between genetic and geographic distances among both the breeding and the non-breeding sites (Z=299, r=0.582, P=0.032, and Z=1721, r=0.620, P=0.022, respectively), suggesting an isolation by distance effect (Fig. 2-5).

The detailed global structure of major features of this pattern were described by the spatial PCA. This analysis revealed significant global structures for both the breeding and non-breeding range (max(t)=0.018, P=0.029 and max(t)=0.019, P =0.024, respectively), i.e. spatially close individuals showed greater genetic similarity to each other than expected in a random spatial distribution (Jombart et al. 2008). No significant local structure was detected (max(t)=0.018, P=0.707 and max(t)=0.023, P=0.317, respectively), i.e. there was no evidence that spatially close individuals exhibited greater genetic differences than random pairs.

Based on the abruptness in the decrease of the sPCA eigenvalues (insets in Figs. 2-6 and 2-7) and their variance and spatial autocorrelation (Appendix Fig. 2-10), we retained the first three global axes (sPC1-3) for interpretation of the breeding ground structure, and the first global axis (sPC1) for the non-breeding grounds.

On the breeding grounds, the sPC1 scores appear to reflect a latitudinal cline across the Alaskan sites, suggesting an isolation by distance pattern (Fig. 2-6). Like the centrally located Kotzebue and Nome breeding populations, individuals at the Siberian site show intermediate scores. However, the Siberian site does not appear to fully fit the latitudinal pattern as its scores appear overall lower than those at the Kotzebue site and more similar to those at the Nome site, which is at a comparable distance but located further south. The second sPC further distinguishes individuals breeding at the Siberia site. As with sPC1, the scores suggest a possible isolation by distance pattern with the sites that are geographically more distant from Siberia (Barrow and the YK-Delta),

appearing genetically more distant than the physically closer sites (Kotzebue and Nome). The third axis (sPC3) appears to reflect genetic differences between individuals at the centrally located Nome site and those at the peripheral Barrow, Siberia and YK-Delta sites, with individuals at the Kotzebue site exhibiting intermediate scores. Similarly to the first breeding sPC, the geographical representation of the non-breeding sPC1 scores reveals a clinal pattern that suggests a latitudinal trend (Fig. 2-7). The overall pattern of latitudinally increasing scores appears to be consistent across the different sex and age classes when these are considered separately (Appendix Fig. 2-12). Non-breeding scores are based on separate analysis and are therefore not directly comparable with breeding ground scores.

2.4.3. Correspondence of Morphology and AFLP data:

All morphometric measurements (culmen, wing and tarsus length) loaded strongly and positively onto the first PC, suggesting that this component best reflects overall size (Appendix Table 2-10). This first PC accounted for 72% of the overall variation (74% when only individuals with genetic data were included). PC2, which might be expected to reveal patterns of shape did not suggest opposite loading patterns for culmen and tarsus and accounted for only 18% of the variation. We therefore retained only the first PC for further analyses.

Individual-based correlation analyses revealed a moderately strong correlation between morphometric PC1 and AFLP sPC1 scores for breeding males but not females (males: r=-0.356, P=0.018, n=35; females: r=0.036, P=0.411, n=41; Fig. 2-8). On the non-breeding grounds, the relationship between morphometric PC1 and AFLP sPC1 scores was significant for both sexes, with a moderate effect for males (r=-0.344, P=0.036, n=28) and a slightly weaker correlation for females (r=-0.280, P=0.019, n=55; Fig. 2-8).

The joint analysis of breeding and non-breeding sites, which was based on 22 AFLP markers and mean site scores, showed a similar latitudinal trend in mean AFLP

PC1 scores across breeding and non-breeding grounds. Additionally, it further supported a correlation between morphometric and AFLP-based PC1 scores (Fig. 2.9). This correlation was strong in both males and females (females: r=0.650, P=0.021, n=10 sites; males: r=0.727, P=0.013, n=9 sites).

2.5. Discussion

2.5.1. mtDNA

The sequenced section of the Western Sandpiper mitochondrial control region was characterized by low nucleotide diversity, with few mutations separating the 40 different haplotypes found among the 120 individuals. The resulting haplotype network showed a star-shape with many rarer haplotypes branching off the common, central haplotype – a pattern that is typical for a population that underwent a recent demographic expansion after experiencing a bottleneck. This suggested demographic expansion scenario was further supported by neutrality tests, all of which were significant.

Star-shaped mitochondrial haplotype networks and/or signals of demographic expansion appear to be common among arctic-breeding shorebirds and have been detected in such species as Semipalmated Sandpipers (Calidris pusilla) Red Knots (Calidris canutus), Temminck’s Stints (Calidris temminckii), Buff-breasted Sandpiper (Tryngites subruficollis), Rock Sandpipers (Calidris ptilocnemis), and Curlew Sandpipers (Calidris ferruginea; Wennerberg 2001; Buehler and Baker 2005; Pruett and Winker 2005; Rönkä et al. 2012; Lounsberry et al. 2013; Miller et al. 2013). In fact, the haplotype networks of some of these species (e.g. the Buff-breasted Sandpiper) show a strikingly similar shape to that of the Western Sandpiper.

Genetic evidence from most if not all of the above-mentioned species suggests they were restricted to one or more refugia during (late) Pleistocene glacial maxima,

when glaciation is presumed to have led to a massive reduction in available breeding habitat (Baker et al. 1994; Wennerberg 2001; Buehler and Baker 2005; Pruett and Winker 2005; Rönkä et al. 2012; Lounsberry et al. 2013). The thus genetically bottlenecked populations appear to have undergone subsequent demographic and range expansion, presumably re-populating wider geographic areas as suitable tundra breeding habitats became more widely available during interglacial periods and/or after the last glacial maximum. In some of these species, low sequence divergences between haplotypes indicate relatively recent events, suggesting expansion likely occurred after the last glacial maximum (e.g. Wennerberg et al. 2002; Lounsberry et al. 2013). Indeed, arctic species in general may frequently be characterized by shallow genetic clades or low diversity as a consequence of range contractions and expansions associated with the most recent glacial period (Hewitt 2004).

Given its arctic breeding range, it is therefore not surprising that the population patterns in Western Sandpiper mitochondrial DNA resemble those seen in many other arctic-breeding shorebirds. Notably, there was only one, very common haplotype with all other haplotypes occurring at much lower frequencies. Furthermore, these rarer haplotypes differed from the main haplotype by only a few substitutions, indicating short evolutionary time-frames. Overall, the resulting star-shaped pattern, the apparent absence of additional maternal lineages and the low level of sequence divergence between haplotypes strongly suggest that Western Sandpipers expanded from a single refugium, likely after the last glacial maximum. There is also a possibility that a selective sweep affecting mitochondrial genes contributed to the observed pattern. While the control region itself is non-coding, it is linked to all other mitochondrial genes and could therefore theoretically be affected by such a sweep (Baker and Marshall 1997).

The Enríquez-Paredes et al. (2012) study, which involved a single non-breeding area in Western Mexico, reached similar conclusions with regard to a historical population expansion. It is of note, however, that their data indicated the presence of a second lineage whose haplotypes were completely absent from our data, despite our study’s large geographic coverage. This second lineage raises the possibility of a second

Pleistocene refugial population, whose extant members may occur at extremely low frequencies across the breeding grounds, or may breed outside the sampled geographic areas. We did not, for instance, sample St. Lawrence Island or the southernmost portion of the breeding range (including the Aleutian islands), and our data only includes one Siberian site.

A possible consequence of a recent demographic expansion is the masking of population structure in largely neutral genetic markers. Population growth tends to counter-act the lineage-eliminating and -sorting effects of genetic drift (Avise 2000). As a consequence, expanding populations experience a delay in reaching genetic equilibrium (Rogers 1995), which essentially obscures the degree of isolation between extant breeding populations. Indeed, the Western Sandpiper mitochondrial control region indicated very little geographical structure. In line with the low nucleotide divergence among haplotypes and the absence of any additional lineages in our data, the ΦST values in our study indicated a complete lack of phylogeographic structure across both the breeding and the non-breeding range. In contrast, significant ΦST values in the Mexican study indicating an uneven temporal and spatial distribution among habitat types, suggested possible differences in migratory behaviours between individuals belonging to the two different maternal lineages detected in this study. Such a difference in migratory strategy was also suggested by differences in the geographical distribution of the two lineages across non-breeding sites within Mexico (Enríquez and Fernandez 2010). Aside from the absence of the second lineage, the haplotype networks derived in our and the Enríquez-Paredes et al. (2012) study showed similar shapes. Additionally, the central, most prevalent haplotype was found at comparable overall frequency. It occurred throughout all sampled sites and while its frequency appeared to vary across sites to some degree, no clear geographical division or cline was apparent. Nevertheless, we found

moderately low but significant frequency-based FST values across breeding and non- breeding sites which indicated some geographic structure on the basis of haplotype frequencies. Frequency- rather than distance-based measures of population divergence can sometimes more accurately represent levels of genetic divergence, particularly if the

underlying structure is very recent and haplotypes are closely related (O'Corry-Crowe et al. 1997). Caution is warranted in interpreting these values, however, as the existence of many rare haplotypes in combination with low to moderate sample sizes is potentially problematic due to simple stochastic effects, i.e. even though haplotypes may exist in a population we may not detect them due to their rarity. The latter problem can be addressed by reducing the number of haplotypes e.g. by only including transversions in the analysis, grouping related haplotypes, or excluding rare haplotypes (Stepien 1999). Given the low number of transversions and the star-shaped haplotype network, only the last option was possible for our analysis and did not appear to change the overall nature of the results. However, given the nature of our data, even the reduced dataset still included haplotypes that were relatively rare. Furthermore, the apparent presence of homoplasy creates an additional level of uncertainty in the estimation of genetic divergence among sites. Hence, while there is some evidence for frequency-based structure, it should be interpreted with caution and is likely to be an imprecise estimate of the actual degree of divergence in this genetic marker.

Moderately low levels of population genetic structure and a lack of phylogeographic sorting across large portions of the species’ breeding and non-breeding range could indicate ongoing gene flow. Alternatively, and as suggested above,

insufficient evolutionary time may have passed for lineage sorting to occur and FST values may not necessarily reflect current levels of gene flow accurately (Whitlock and McCauley 1999; Bulgin et al. 2003). Given the strong evidence of an evolutionarily recent bottleneck and subsequent expansion, any divergence would be expected to be recent and consequently would likely be difficult to detect in genetic markers that are not under selection or closely linked to genes under selection, as may be the case for the mitochondrial control region in this species. Thus the current genetic patterns could simply reflect a combination of unsorted ancestral haplotypes and many new, but rare haplotypes as consequences of the recent demographic expansion. A combination of both factors may also explain the observed patterns. Ultimately, the low, but significant

frequency-based differences suggest the species is not completely panmictic and thus may indicate beginning divergence.

Low levels of divergence, and even a complete absence of geographic structure (on the basis of mitochondrial DNA sequences), are frequently found among shorebird species that underwent recent demographic or range expansions. For instance, the sister species of the Western Sandpiper, the Semipalmated Sandpiper, shows comparably low levels of intraspecific divergence in the control region, despite its geographically more extensive breeding range (Miller et al. 2013). As in the Western Sandpiper, among-site differences were based on haplotype frequencies rather than the presence of distinct maternal lineages. Other species such as the Temminck’s Stint, Buff-breasted Sandpiper, Common Sandpiper (Actitis hypoleucos), Ruff (Philomachus pugnax), and Mountain Plover show no significant population structuring on the basis of mitochondrial DNA (Oyler-McCance et al. 2005; Rönkä et al. 2008; Zink et al. 2008; Rönkä et al. 2012; Verkuil et al. 2012; Lounsberry et al. 2013). Similarly, no clear geographic structure emerges for Turnstones despite global sampling (Wenink et al. 1994), while in Curlew Sandpipers structure is limited to possible low levels of divergence among individuals using different migratory flyways (Wennerberg 2001), and Red Knots exhibit levels of divergence even among subspecies (Buehler and Baker 2005). In several species, the observed low levels of mitochondrial structure stand in marked contrast to notable geographic variation in phenotypic traits such as morphology, plumage or migratory behaviour, e.g. in Red Knots, White-rumped Sandpipers, Ruffs, Redshanks and Semipalmated Sandpipers (Wennerberg et al. 2002; Buehler and Baker 2005; Ottvall et al. 2005; Verkuil et al. 2012; Miller et al. 2013). The seeming mismatch between phenotypic and genetic variation in these species and in the Western Sandpiper might indicate that the expression of the phenotypic characters in question is largely driven by environmental effects or environment-gene interactions. However, the patterns observed in genetic markers not experiencing selection, as may be the case for the mitochondrial control region, do not necessarily provide an accurate portrayal of the level of divergence in other parts of the genome, particularly those under selection (Pruett and Winker 2008).

In fact, there can be large discrepancies among mitochondrial, nuclear and phenotypic measures of divergence (Humphries and Winker 2011). It is therefore possible, that the respective phenotypic differences have a genetic background and selective pressures in conjunction with the recent range-expansion led to an evolutionary divergence in these traits. If the divergence is sufficiently recent, it may not yet have led to lineage sorting in genetic markers not driven by selection and hence would not be evident in patterns derived from such markers (Bensch et al. 1999; Bulgin et al. 2003; Buehler and Baker 2005). While mtDNA can be a useful marker for population genetic studies due to its high mutation rate, low effective population size, and lack of recombination, it does not universally provide a better resolution than nuclear markers, particularly if divergence estimates are based on several nuclear markers (Larsson et al. 2009; Humphries and Winker 2011). Due to its smaller effective population size, bottleneck and subsequent expansion events are likely to have a greater effect on mtDNA. Bottlenecks can lead to the random fixation of one particular haplotype, which can then be carried along as the population expands again (Humphries and Winker 2011). Nuclear genes should be affected to a lesser degree by such demographic changes and therefore better represent actual levels of divergence in these cases (Humphries and Winker 2011). Thus, in cases of recent range and population expansion after a demographic bottleneck with recent and likely still evolving population structure, nuclear markers, ideally those evolving at even higher rates than the mtDNA control region, may be required to adequately resolve population structure (Baker et al. 1994).

That geographic variation in shorebird morphology and migratory behaviour can have a population genetic background is suggested by species such as the Dunlin or Rock Sandpiper in which phenotypic traits and mtDNA patterns show at least a partial congruence (Wennerberg et al. 1999; Pruett and Winker 2005). However, divergence times in these species are typically older (i.e. predating the last glacial maximum), again suggesting that sufficient time needs to have passed before such patterns become evident in mitochondrial markers.

2.5.2. AFLP:

Like the mitochondrial control region sequences, the AFLP data showed moderately low, but significant population structure across both the breeding and non- breeding range with overall FST values being lower than the corresponding mitochondrial

ΦST values. Correspondingly, fewer pairwise comparisons were significant, with results nevertheless indicating significant differences in marker frequencies across many

sites/sex comparisons. While the absolute values of the overall FST were low, divergence appears reasonably high on a relative scale, when the relatively small distances among breeding populations are taken into account.

Across the breeding range, none of the AFLP markers we analyzed were fixed (present in all individuals in one and completely absent in another population). Similarly, there were no fixed differences among non-breeding sites within each sex. The only two, seemingly fixed differences involved inter-sex comparisons across sites with very small samples sizes (California females, n=4, vs. Yucatan males, n =6, and Panama males, n =4). They are therefore unlikely to truly represent fixed differences among populations, particularly since the respective bands were present at all of these sites when both sexes were considered. Only one marker occurred as a private allele in our dataset (i.e. was exclusively found in one population) and was represented at very low frequency (11%) at the Barrow site.

Low levels of differentiation in nuclear genetic markers among populations and even subspecies appear to be the norm rather than the exception for many shorebird species. Some species such as the Black-tailed Godwit (Limosa limosa), Snowy Plover (Charadrius nivosus), and closely related Kentish Plover (Charadrius alexandrinus) show considerable differentiation between continents or continental and island-based populations, but display little or no geographic structure within continental ranges (Funk et al. 2007; Küpper et al. 2012; Trimbos et al. 2014). Many species exhibit moderately low levels of divergence, e.g. Dunlin, Temminck’s Stints, Redshanks or Bristle-thighed Curlews (Numenius tahitiensis; Wennerberg 2001; Ottvall et al. 2005; Marthinsen et al.

2007; Rönkä et al. 2012; Miller et al. 2015; Sonsthagen et al. 2015). Whereas a fair number of species appear to show no clear geographic structure in the surveyed nuclear markers, among them New Zealand Dotterels (Charadrius obscurus), Buff-breasted Sandpipers, Common Sandpipers, Curlew Sandpipers, Ruffs and Semipalmated Sandpipers (Wennerberg 2001; Verkuil et al. 2012; Barth et al. 2013; Lounsberry et al. 2013; Miller et al. 2013). Similar to what has been suggested for many of these species, and in accordance with the mitochondrial DNA, the low levels of geographic structure in the Western Sandpiper AFLP data suggest ongoing, though not unlimited, gene flow across the species’ range or very recent divergence with incomplete lineage sorting.

In contrast to the mitochondrial sequence results, however, the nuclear marker analysis appeared to show broader geographic patterns in addition to indicating a limited degree of genetic heterogeneity among sites. More specifically, Mantel tests suggested a substantial effect of geographic distance on genetic differences among sites (large effect sizes). This isolation by distance pattern was also evident in the geographical patterns revealed by the spatial Principal Components Analyses. Interestingly, scores for the first sPC indicated a latitudinal cline across breeding sites. Additional, but weaker structure, also suggested that some of the genetic variation may be explained by divergences of the easternmost and central breeding sites respectively. Again, the data suggested gradual patterns rather than an abrupt transition. It is possible that founder effects play a role for the easternmost site in Siberia, reflecting recent colonization by a limited number of individuals. However, blood samples from the Siberian site were stored in a different medium (ethanol) than samples from Alaskan sites (lysis buffer), which could potentially have led to differences in AFLP band patterns, e.g. through greater degradation of DNA. While we excluded samples that showed signs of degradation in an Agarose electrophoresis gel or visibly deviant band patterns in the AFLP, we cannot exclude the possibility of more subtle effects that would not have been easily detected. However, systematic effects are unlikely except for overall fewer high molecular weight fragments as a result of degradation for which we found no evidence (data not shown).

Therefore, further study with higher quality samples would be required to confirm or repute a possible divergence of populations breeding in Siberia. Given its central location and spatial proximity to other breeding sites, the Nome site seems an unlikely candidate for divergence. Possibly, the data, rather than indicating isolation of the Nome site, reflects the gradual divergence of peripheral sites from a centrally located source population. However, this hypothesis would have to be further tested by including additional data and determining genetic diversity in nuclear markers at central vs. peripherally located sites.

The first, latitudinal pattern on the breeding grounds was mirrored by a latitudinal structure on the non-breeding grounds, while the latter two had no clearly evident correspondences across the non-breeding range. The latitudinal increase in non-breeding sPC1 scores was consistent across all four sex/age groups and a joint PCA of breeding and non-breeding samples indicated corresponding latitudinal clines across the two ranges. Such latitudinal clines were what we expected to see if the non-breeding variation in morphological traits had a population genetic component.

2.5.3. Correspondence of Morphology and AFLP data:

Our data did, in fact, indicate some correlation between body size and genetic variation, both on an individual and a site based level. Average site scores suggested a strong correlation. On an individual level, however, effects were weak to moderate with breeding females showing no detectable correlation between AFLP-based sPC and body size scores. The latter could be the result of measurement inaccuracies in the morphological data, particularly as site-based comparisons that included breeding and non-breeding data showed significant relationships for both sexes. Alternatively, it might reflect higher dispersal tendencies (i.e. lower levels of philopatry) among females, in conjunction with environment-gene interactions.

In the event of a recent, selection-driven divergence, genetic markers not immediately under selection may be only loosely associated with morphological

characteristics, provided they are not tightly linked to the quantitative trait loci influencing these traits (Bensch et al. 1999; Humphries and Winker 2011). Indeed, in some shorebird species geographic variation in morphology could not be linked to a corresponding divergence in nuclear markers (Wennerberg et al. 2002; Verkuil et al. 2012). An apparent correspondence between nuclear genetic markers and morphological traits has been found in a few species, though, suggesting a possible population genetic background for the geographic variation in morphology (Wennerberg 2001; Ottvall et al. 2005). On an individual level, the association between random genetic markers and morphological traits might be even less pronounced, which could explain the weak to moderate correlations found in our individual-based analyses. On the other hand, the weaker correspondence at the individual than at the site level may indicate that environmental factors play a considerable role in shaping morphology in Western Sandpipers, leading to a substantial amount of variation within each population. That environmental influences on morphology can be substantial has been shown in Red- winged Blackbirds (Agelaius phoeniceus; James 1983). Some of the variation in morphology is likely also attributable to measurement imprecision and even to slight differences in the way different researchers measured these birds. On the whole, our morphological data has to be regarded with a great deal of caution as it was collected by many different researchers and tarsus and wing measurements had to be converted in some instances, adding noise and possible sources of bias to the data. A pilot study suggested there may be biases and consistent differences even among experienced banders (unpublished data). Additionally, differences in wing wear influence wing length measurements to some degree. However, the fact that data at most sites was collected by several or even many different researchers and that some researchers collected morphological data across many different sites should reduce the effect of potential biases (though not noise) on geographic patterns. Furthermore, while bias may affect individual sites, the large-scale pattern across both the breeding and non-breeding range is unlikely to have arisen from banding bias alone. Finally, the implied latitudinal patterns in body size corresponded to expected trends: on the non-breeding grounds where previous data suggested an increase in body size and/or wing and culmen length

from northern to southern sites (Nebel 2005; O'Hara et al. 2006; Stein et al. 2008; Franks 2012), as well as on the breeding grounds where male song frequencies increased with latitude, suggesting a corresponding decrease in body size (Chapters 3 and 4). Overall, our data suggests that some of the variation in overall body size may have a population genetic background. Nevertheless, the many possible sources of bias and noise add a substantial degree of uncertainty about the strength of the relationship between genetic and morphological traits and thus their role in driving migratory patterns in this species.

In conjunction, the nuclear and morphological data suggest a possible ‘chain’-like migration system, one that is based on a gradual pattern of isolation by distance rather than on clear breaks between distinct populations. At the same time, the genetic population structure was very low and its association with the variation in body size appeared to be moderate. Despite similar latitudinal trends, we also did not see clear and consistent correspondences between breeding and non-breeding sites that might be expected to match based on their relative geographical position within the species’ latitudinal range. In fact, pairwise FST values indicated significant differences between some southern, central and northern site pairs. In part, this might be due to not having sampled the exact matching populations, however it is more likely an indicator for relatively weak migratory connectivity and mixing of different breeding populations at the respective non-breeding sites. The joint breeding and non-breeding PCA, while suggesting similar latitudinal clines, also did not reveal obvious, clear-cut matches between southern, central and northern breeding and non-breeding sites. The southernmost non-breeding sites also had larger mean scores than any breeding site. The latter may reflect the absence of the southernmost areas of the breeding range in our dataset, from which we were unable to obtain samples. Given all of the above, it appears likely that environmental effects also play a defining role in determining an individual’s morphology. Similarly, where an individual spends its non-breeding season may ultimately be determined by a combination of its genetic background and additional factors such as its morphological traits, physiological condition, age (see Nebel et al.

2002) and even an element of chance. In line with this hypothesis, two of our overall FST

estimates indicated that the genetic differences among sites might be slightly lower among non-breeding than breeding sites. The third, substantially higher estimate, is unlikely to be accurate, given its deviation from all other estimates and may be attributable to a less reliable allele frequency estimate. A weaker structure on the non- breeding grounds might also account for the lack of additional levels of structure in the sPCA that would correspond to those found on the breeding grounds. However, other explanations such as low sample sizes and incomplete sampling are also plausible. Intra- seasonal movements have been suggested for the sister species, the Semipalmated Sandpiper, and local congeners (likely including Western Sandpipers) on Caribbean non- breeding grounds (Rice et al. 2007). Such movements would further indicate a certain degree of spatial flexibility and suggest that behavioural decisions play at least some role in non-breeding site selection. On the other hand, non-breeding site fidelity appears to be high in Western Sandpipers at other sites (Smith and Stiles 1979; Fernández et al. 2003; Fernández et al. 2004; O'Hara et al. 2007), with little apparent movement even between regional sites (O’Hara personal comm. as cited by Franks et al. 2014). This does not, however, preclude behaviourally-mediated plasticity in the initial site selection nor a change in site selection after the first year, which presumably mediates the age-dependent distribution patterns observed in this species (Nebel et al. 2002). Ultimately, even a relatively plastic migratory program, that incorporates aspects of learning and behavioural decision-making, may lead to broad-scale connectivity patterns on the non- breeding grounds. This may particularly be the case where morphology acts as a link between genetic differences and migratory strategies and/or site preferences.

Our data suggests that an individual’s genetic background plays some role in determining how far south it migrates (either directly or indirectly through an effect on morphological characteristics). Nevertheless, a scan of thousands of potential AFLP markers revealed no fixed differences between female juveniles at a far north and those at a far south non-breeding site along the Pacific Coast, despite their presumed differences in migratory and life history strategies. Similarly, no fixed differences between these juvenile females and those at an Atlantic Coast non-breeding site were apparent in our

scan. We therefore found no evidence that the presumed differences in migratory direction between Pacific and Atlantic non-breeding populations relate to fixed genetic differences among these populations.

2.6. Synthesis

In sum, our data suggest a ‘chain’-like migration system, but neither indicate strong migratory connectivity nor high levels of divergence among breeding or non- breeding populations. Given the small and recent genetic differences among breeding populations, it does not appear likely that the morphological and life history differences among non-breeding sites can be fully explained by population genetic differences. While genetic structure was too low to directly attempt an assignment of non-breeding individuals to their breeding origins, neither the mtDNA nor the AFLP data provided any evidence for strong migratory connectivity. This lack of support for a close link among specific breeding and non-breeding sites echoes the results from a stable isotope based study, which found little evidence for strong connectivity (Franks et al. 2012). Only males from the Nome and YK-Delta breeding sites had higher likelihoods of having migrated to particular non-breeding areas than would be expected by chance under a population-size based model. Given these results, even a somewhat flexible, ‘chain’-like migration system as suggested by our study may seem surprising. However, the detection of migratory connectivity between breeding and non-breeding areas through isotope signatures in feathers may, on the one hand, be severely complicated by flexible molt schedules, that appear to include a large proportion of molt-migration strategies (Franks 2012). On the other hand, our data does not rule out that substantial mixing occurs; rather it suggests that migratory tendencies likely are influenced by population genetic origins to some degree.

Given the evidence from our study along with results from earlier publications, it appears likely that Western Sandpipers have undergone an evolutionarily recent range and population expansion that has resulted in divergence patterns that are primarily

isolation by distance driven. Furthermore, what data are available to date, suggest their migration could be explained by a relatively weak genetic migratory program. Such a program may allow for a considerable amount of phenotypic plasticity that provides room for behavioural decision-making. It is possible that this flexibility also acts in reverse, i.e. allows individuals to migrate to and breed at breeding sites that are distinct from their natal site. Such dispersal would promote mixing between populations with the resulting higher levels of gene flow impeding lineage sorting and divergence. Genetically controlled migration tendencies could constrain the magnitude of the dispersal distances, creating a gradual pattern that would be consistent with the observed isolation by distance patterns and overall low genetic population structure in this species. Rather than being under direct endogenous control, migratory behaviour might also be indirectly linked to an individual’s population origin, e.g. if populations differ in certain morphological characteristics (e.g. size or wing shape) that convey a propensity to migrate shorter or larger distances or influence non-breeding site choice. Factors that may be driving morphology-mediated site choices include food burial depth and predation danger (Nebel 2005; Nebel and Ydenberg 2005; Mathot et al. 2007). If this is the case, present-day fluctuations in raptor numbers, along with climatic and habitat changes carry the potential to drive adaptive changes that may not (yet) be evident in the genetic data. If both breeding and non-breeding site choices are mediated through morphology, morphology driven migratory patterns might result in genetic divergence among populations (rather than being a consequence of this). In this case, migration patterns could be primarily or entirely based on phenotypic plasticity, but would result in body size based segregation on the breeding grounds, and thus in assortative mating. If body size has a significant genetic component, offspring could be expected to show similar migration patterns which over time could lead to genetic segregation.

One phenomenon that clearly merits further investigation is the absence of the second mitochondrial lineage in our study. So far, this lineage has only been found at Mexican non-breeding sites with an uneven temporal and spatial distribution (Enríquez and Fernandez 2010; Enríquez-Paredes et al. 2012). A survey of a major migratory

stopover area on the Pacific Coast, the Fraser River Delta (FRD), also failed to detect this lineage in the sampled individuals (Hemmera Envirochem. Inc. 2014). The absence of this lineage in our breeding ground data suggests that these individuals may originate from an as yet unexplored breeding population that winters primarily or exclusively in Mexico. Alternatively, individuals that possess these haplotypes may exist at such low frequencies within the studied populations that we failed to detect them, but may migrate at higher frequencies to specific non-breeding areas, increasing their detectability at these sites. The Mexico data further suggested differences among Atlantic and Pacific Coast non-breeding populations, possibly indicating a stronger presence of the second lineage on the Atlantic Coast (Enríquez and Fernandez 2010). This might explain the apparent absence (or low frequency) of this lineage at the Pacific Coast stopover site, as Atlantic populations might utilize a more easterly migration route. While a stronger divergence of Atlantic Coast populations was not immediately apparent in our data, possibly due to limited sample and site numbers, further investigation is needed to clarify to what degree population genetic origins affect migratory direction in this species.

Finally, we cannot exclude the possibility that levels of divergence may be higher for a small portion of the genome containing genes under strong selection for local adaptation which may have been missed in our study. Genetic divergence at largely neutral loci may initially remain low under certain scenarios of selection-driven divergence (Emelianov et al. 2004; Thibert-Plante and Hendry 2009, 2010) and may not reflect differences in morphology or migratory direction (Bensch et al. 1999). Differentiation in traits associated with migration in particular appears to typically involve selection on a relatively small part of the genome with traits generally showing at most a loose correlation with the overall genetic population structure (Liedvogel et al. 2011). However, even if parts of the Western Sandpiper genome are under divergent selection, it seems unlikely that the affected genes would be tied to rigid, population- specific migratory behaviours, given the overall low evidence for strong migratory connectivity.

Overall, the current data suggest flexibility in migratory strategies, but also indicates that there may be limits to this flexibility on a larger scale. Until further data are available, we therefore suggest that efficient conservation strategies might involve focussing on the protection of major breeding, migratory stop-over and non-breeding areas that sustain large numbers of Western Sandpipers. At the same time, an effort should be made to include sites located at different latitudes across the species’ breeding and non-breeding range to ensure the viability of different migratory and life history strategies. Light-level geolocators and next-generation sequencing technologies could be utilized in future studies to obtain a more detailed understanding of Western Sandpiper migration strategies, which may in turn enable more finely tuned conservation strategies.

2.7. References

Avise, J. C. 2000. Phylogeography: the history and formation of species. Harvard University Press, Cambridge, MA.

Baker, A. J., and H. D. Marshall. 1997. Mitochondrial Control Region Sequences as Tools for Understanding Evolution. Pages 51-82 in Avian Molecular Evolution and Systematics (D. P. Mindell, Ed.). Academic Press, San Diego, CA.

Baker, A. J., T. Piersma, and L. Rosenmeier. 1994. Unraveling the Intraspecific Phylogeography of Knots Calidris canutus: a Progress Report on the Search for Genetic Markers. Journal Für Ornithologie 135:599-608.

Barr, K. R., B. E. Kus, K. L. Preston, S. Howell, E. Perkins, and A. G. Vandergast. 2015. Habitat fragmentation in coastal southern California disrupts genetic connectivity in the cactus wren (Campylorhynchus brunneicapillus). Molecular Ecology 24:2349-2363.Barth, J. M. I., M. Matschiner, and B. C. Robertson. 2013. Phylogenetic Position and Subspecies Divergence of the Endangered New Zealand Dotterel (Charadrius obscurus). PLoS ONE 8.

Bensch, S. 1999. Is the range size of migratory birds constrained by their migratory program? Journal of Biogeography 26:1225-1235.

Bensch, S., T. Andersson, and S. Åkesson. 1999. Morphological and molecular variation across a migratory divide in willow warblers, Phylloscopus trochilus. Evolution 53:1925-1935.

Bensch, S., M. Grahn, N. Müller, L. Gay, and S. Åkesson. 2009. Genetic, morphological, and feather isotope variation of migratory willow warblers show gradual divergence in a ring. Molecular Ecology 18:3087-3096.

Bonin, A., D. Ehrich, and S. Manel. 2007. Statistical analysis of amplified fragment length polymorphism data: a toolbox for molecular ecologists and evolutionists. Molecular Ecology 16:3737-3758.

Buehler, D. M., and A. J. Baker. 2005. Population divergence times and historical demography in Red Knots and Dunlins. Condor 107:497-513.

Bulgin, N. L., H. L. Gibbs, P. Vickery, and A. J. Baker. 2003. Ancestral polymorphisms in genetic markers obscure detection of evolutionarily distinct populations in the endangered Florida grasshopper sparrow (Ammodramus savannarum floridanus). Molecular Ecology 12:831-844.

Elner, R. W., and D. A. Seaman. 2003. Calidrid conservation: unrequited needs. Wader Study Group Bulletin 100.

Emelianov, I., F. Marec, and J. Mallet. 2004. Genomic evidence for divergence with gene flow in host races of the larch budmoth. Proceedings of the Royal Society B- Biological Sciences 271:97-105.

Engelmoer, M., and C. S. Roselaar. 1998. Geographical Variation in Waders. Kluwer Academic Publishers, Dordrecht.

Enríquez, L. M., and G. Fernandez. 2010. Genetic identity and population structure of nonbreeding Western Sandpiper in Mexico. Poster presented at the COS/AOU/SCO 2010 Joint Meeting, San Diego, CA.

Esler, D. 2000. Applying metapopulation theory to conservation of migratory birds. Conservation Biology 14:366-372.

Excoffier, L., and H. E. L. Lischer. 2010. Arlequin suite ver 3.5: A new series of programs to perform population genetics analyses under Linux and Windows. Molecular Ecology Resources 10:564-567.

Fernández, G., H. de la Cueva, N. Warnock, and D. B. Lank. 2003. Apparent survival rates of Western Sandpiper (Calidris mauri) wintering in northwest Baja California, Mexico. Auk 120:55-61.

Fernández, G., P. D. O'Hara, and D. B. Lank. 2004. Tropical and subtropical Western Sandpipers (Calidris mauri) differ in life history strategies. Ornitologia Neotropical 15:385-394.

Formia, A., A. C. Broderick, F. Glen, B. J. Godley, G. C. Hays, and M. W. Bruford. 2007. Genetic composition of the Ascension Island green turtle rookery based on mitochondrial DNA: implications for sampling and diversity. Endangered Species Research 3:145-158.

Funk, W. C., T. D. Mullins, and S. M. Haig. 2007. Conservation genetics of snowy plovers (Charadrius alexandrinus) in the Western Hemisphere: population genetic structure and delineation of subspecies. Conservation Genetics 8:1287-1309.

Gaudeul, M., G. Rouhan, M. F. Gardner, and P. M. Hollingsworth. 2012. AFLP markers provide insights into the evolutionary relationships and diversification of New Caledonian Araucaria species (Araucariaceae). American Journal of Botany 99:68-81.

Gratto-Trevor, C., R. I. G. Morrison, D. Mizrahi, D. B. Lank, P. Hicklin, and A. L. Spaans. 2012. Migratory Connectivity of Semipalmated Sandpipers: Winter Distribution and Migration Routes of Breeding Populations. Waterbirds 35:83-95.

Haig, S. M., W. M. Bronaugh, R. S. Crowhurst, J. D'Elia, C. A. Eagles-Smith, C. W. Epps, B. Knaus, M. P. Miller, M. L. Moses, S. Oyler-McCance, W. D. Robinson, and B. Sidlauskas. 2011. Genetic Applications in Avian Conservation. Auk 128:205-229.

Haig, S. M., C. L. GrattoTrevor, T. D. Mullins, and M. A. Colwell. 1997. Population identification of western hemisphere shorebirds throughout the annual cycle. Molecular Ecology 6:413-427.

Hall, T. A. 1999. BioEdit: a user-friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. Nucl. Acids. Symp. Ser. 41:95-98.

Hemmera Envirochem. Inc. 2014. Roberts Bank Terminal 2 Technical Data Report. Coastal Birds. Genetics of Western Sandpipers Using the Fraser River Estuary during Northward Migration. 40 pages. Prepared by Hemmera Environchem. Inc. for Port Metro Vancouver.

Henningsson, S. S., and T. Alerstam. 2006. Implications of migratory connectivity for species' ranges and subspeciation of arctic shorebirds. Ardea 94:499-509.

Hewitt, G. M. 2004. Genetic consequences of climatic oscillations in the Quaternary. Philosophical Transactions of the Royal Society B-Biological Sciences 359:183-195.

Holsinger, K. E., P. O. Lewis, and D. K. Dey. 2002. A Bayesian approach to inferring population structure from dominant markers. Molecular Ecology 11:1157-1164.

Humphries, E. M., and K. Winker. 2011. Discord reigns among nuclear, mitochondrial and phenotypic estimates of divergence in nine lineages of trans-Beringian birds. Molecular Ecology 20:573-583.

Irwin, D. E. 2009. Incipient ring speciation revealed by a migratory divide. Molecular Ecology 18:2923-2925.

Irwin, D. E., and J. H. Irwin. 2005. Siberian migratory divides: the role of seasonal migration in speciation. Pages 27-40 in Birds of Two Worlds: The Ecology and Evolution of Migration (R. Greenberg, and P. P. Marra, Eds.). Johns Hopkins University Press, Baltimore.

James, F. C. 1983. Environmental Component of Morphological Differentiation in Birds. Science 221:184-186.

Jensen, J. L., A. J. Bohonak, and S. T. Kelley. 2005. Isolation by distance, web service. BMC Genetics 6:13. v.13.23 http://ibdws.sdsu.edu. .

Jombart, T. 2008. adegenet: a R package for the multivariate analysis of genetic markers. Bioinformatics 24:1403-1405.

Jombart, T., S. Devillard, A. B. Dufour, and D. Pontier. 2008. Revealing cryptic spatial patterns in genetic variability by a new multivariate method. Heredity 101:92-103.

Jombart, T., D. Pontier, and A. B. Dufour. 2009. Genetic markers in the playground of multivariate analysis. Heredity 102:330-341.

Kraaijeveld, K. 2008. Non-breeding habitat preference affects ecological speciation in migratory waders. Naturwissenschaften 95:347-354.

Küpper, C., S. V. Edwards, A. Kosztolanyi, M. Alrashidi, T. Burke, P. Herrmann, A. Arguelles-Tico, J. A. Amat, M. Amezian, A. Rocha, H. Hotker, A. Ivanov, J. Chernicko, and T. Szekely. 2012. High gene flow on a continental scale in the polyandrous Kentish plover Charadrius alexandrinus. Molecular Ecology 21:5864-5879.

Larsson, L. C., J. Charlier, L. Laikre, and N. Ryman. 2009. Statistical power for detecting genetic divergence-organelle versus nuclear markers. Conservation Genetics 10:1255-1264.

Librado, P., and J. Rozas. 2009. DnaSP v5: a software for comprehensive analysis of DNA polymorphism data. Bioinformatics 25:1451-1452.

Liedvogel, M., S. Åkesson, and S. Bensch. 2011. The genetics of migration on the move. Trends in Ecology & Evolution 26:561-569.

Liu, Y., I. Keller, and G. Heckel. 2012. Breeding site fidelity and winter admixture in a long-distance migrant, the tufted duck (Aythya fuligula). Heredity 109:108-116.

Longmire, J., M. Maltbie, and R. Baker. 1997. Use of "Lysis Buffer" in DNA isolation and its implication for museum collections. Occasional paper. Museum of Texas Tech University. Number 163.

Lynch, M., and B. G. Milligan. 1994. Analysis of population genetic structure with RAPD markers. Molecular Ecology 3:91-99.

Marthinsen, G., L. Wennerberg, and J. T. Lifjeld. 2007. Phylogeography and subspecies taxonomy of dunlins (Calidris alpina) in western Palearctic analysed by DNA microsatellites and amplified fragment length polymorphism markers. Biological Journal of the Linnean Society 92:713-726.

Mathot, K. J., B. D. Smith, and R. W. Elner. 2007. Latitudinal clines in food distribution correlate with differential migration in the Western Sandpiper. Ecology 88:781-791.

Miller, M. P. 1997. Tools for Population Genetic Analyses (TFPGA). A Windows® program for the analysis of allozyme and molecular population genetic data. Computer software distributed by author.

Miller, M. P., C. Gratto-Trevor, S. M. Haig, D. S. Mizrahi, M. M. Mitchell, and T. D. Mullins. 2013. Population Genetics and Evaluation of Genetic Evidence for Subspecies in the Semipalmated Sandpiper (Calidris pusilla). Waterbirds 36:166-178.

Miller, M. P., S. M. Haig, T. D. Mullins, L. Z. Ruan, B. Casler, A. Dondua, H. R. Gates, J. M. Johnson, S. Kendall, P. S. Tomkovich, D. Tracy, O. P. Valchuk, and R. B. Lanctot. 2015. Intercontinental genetic structure and gene flow in Dunlin (Calidris alpina), a potential vector of avian influenza. Evolutionary Applications 8:149-171.

Nebel, S. 2005. Latitudinal clines in bill length and sex ratio in a migratory shorebird: a case of resource partitioning? Acta Oecologica-International Journal of Ecology 28:33-38.

Nebel, S., and R. Ydenberg. 2005. Differential predator escape performance contributes to a latitudinal sex ratio cline in a migratory shorebird. Behavioral Ecology and Sociobiology 59:44-50.

O'Corry-Crowe, G. M., R. S. Suydam, A. Rosenberg, K. J. Frost, and A. E. Dizon. 1997. Phylogeography, population structure and dispersal patterns of the beluga whale Delphinapterus leucas in the western Nearctic revealed by mitochondrial DNA. Molecular Ecology 6:955-970.

O'Hara, P. D. 2003. The role of feather wear in alternative life history strategies of a long-distance migratory shorebird, the western sandpiper (Calidris mauri). Ph.D. thesis, Simon Fraser University, Burnaby.

O'Hara, P. D., G. Fernandez, F. Becerril, H. de la Cueva, and D. B. Lank. 2005. Life history varies with migratory distance in Western Sandpipers Calidris mauri. Journal of Avian Biology 36:191-202.

O'Hara, P. D., G. Fernandez, B. Haase, H. de la Cueva, and D. B. Lank. 2006. Differential migration in Western Sandpipers with respect to body size and wing length. Condor 108:225-232.

O'Hara, P. D., B. J. M. Haase, R. W. Elner, B. D. Smith, and J. K. Kenyon. 2007. Are population dynamics of shorebirds affected by El Nino/Southern Oscillation (ENSO) while on their non-breeding grounds in Ecuador? Estuarine Coastal and Shelf Science 74:96-108.

Ottvall, R., J. Hoglund, S. Bensch, and K. Larsson. 2005. Population differentiation in the redshank (Tringa totanus) as revealed by mitochondrial DNA and amplified fragment length polymorphism markers. Conservation Genetics 6:321-331.

Oyler-McCance, S. J., J. St John, F. L. Knopf, and T. W. Quinn. 2005. Population genetic analysis of Mountain Plover using mitochondrial DNA sequence data. Condor 107:353-362.

Page, G., and B. Fearis. 1971. Sexing Western Sandpipers by Bill Length. Bird- Banding 42:297-298.

Pruett, C. L., and K. Winker. 2008. Evidence for cryptic northern refugia among high- and temperate-latitude species in Beringia - A response to Stewart and Dalen (2008). Climatic Change 86:23-27.

Pulido, F. 2007. The genetics and evolution of avian migration. Bioscience 57:165-174.

Rice, S. M., J. A. Collazo, M. W. Alldredge, B. A. Harrington, and A. R. Lewis. 2007. Local annual survival and seasonal residency rates of semipalmated sandpipers (Calidris pusilla) in Puerto Rico. Auk 124:1397-1406.

Rogers, A. R. 1995. Genetic Evidence for a Pleistocene Population Explosion. Evolution 49:608-615.

Rönkä, A., L. Kvist, J. Karvonen, K. Koivula, V. M. Pakanen, D. Schamel, and D. M. Tracy. 2008. Population genetic structure in the Temminck's stint Calidris temminckii, with an emphasis on Fennoscandian populations. Conservation Genetics 9:29-37.

Rouger, R., and A. S. Jump. 2015. Fine-scale spatial genetic structure across a strong environmental gradient in the saltmarsh plant Puccinellia maritima. Evolutionary Ecology 29:609-623.

Schlüter, P. M., and S. A. Harris. 2006. Analysis of multilocus fingerprinting data sets containing missing data. Molecular Ecology Notes 6:569-572.

Senner, S. E., G. C. West, and D. W. Norton. 1981. The Spring Migration of Western Sandpipers and Dunlins in Southcentral Alaska: Numbers, Timing, and Sex- Ratios. Journal of Field Ornithology 52:271-284.

Seutin, G., B. N. White, and P. T. Boag. 1991. Preservation of Avian Blood and Tissue Samples for DNA Analyses. Canadian Journal of Zoology-Revue Canadienne De Zoologie 69:82-90.

Shirk, A. J. 2009. Mountain Goat Genetic Structure, Molecular Diversity, and Gene Flow in the Cascade Range, Washington. M.Sc. Thesis, Western Washington University.

Smith, S. M., and F. G. Stiles. 1979. Banding studies of migrant shorebirds in northwestern Costa Rica. . Pages 41-47 in Shorebirds in marine environments. (F. A. Pitelka, Ed.). Cooper Ornithological Society, Lawrence, Kansas.

Sonsthagen, S., T. Tibbitts, R. Gill, I. Williams, and S. Talbot. 2015. Spatial genetic structure of bristle-thighed curlews (Numenius tahitiensis): breeding area differentiation not reflected on the non-breeding grounds. Conservation Genetics 16:223- 233.

Storey, J. D. 2002. A direct approach to false discovery rates. Journal of the Royal Statistical Society Series B-Statistical Methodology 64:479-498.

Takezaki, N., and M. Nei. 1996. Genetic distances and reconstruction of phylogenetic trees from microsatellite DNA. Genetics 144:389-399.

Teich, I., A. Verga, and M. Balzarini. 2014. Assessing spatial genetic structure from molecular marker data via principal component analyses: A case study in a Prosopis sp. forest. Advances in Bioscience and Biotechnology 5:89-99.

Thibert-Plante, X., and A. P. Hendry. 2009. Five questions on ecological speciation addressed with individual-based simulations. Journal of Evolutionary Biology 22:109-123.

Thibert-Plante, X., and A. P. Hendry. 2010. When can ecological speciation be detected with neutral loci? Molecular Ecology 19:2301-2314.

Toews, D. P. L., and D. E. Irwin. 2008. Cryptic speciation in a Holarctic passerine revealed by genetic and bioacoustic analyses. Molecular Ecology 17:2691-2705.

Vader Wal, E., I. Edye, P. C. Paquet, D. W. Coltman, E. Bayne, R. K. Brook, and J. A. Andrés. 2013. Juxtaposition between host population structures: implications for disease transmission in a sympatric cervid community. Evolutionary Applications 6:1001-1011.

Verkuil, Y. I., T. Piersma, J. Jukema, J. C. E. W. Hooijmeijer, L. Zwarts, and A. J. Baker. 2012. The interplay between habitat availability and population differentiation: a case study on genetic and morphological structure in an inland wader (Charadriiformes). Biological Journal of the Linnean Society 106:641-656.

Vos, P., R. Hogers, M. Bleeker, M. Reijans, T. Vandelee, M. Hornes, A. Frijters, J. Pot, J. Peleman, M. Kuiper, and M. Zabeau. 1995. AFLP - a new technique for DNA fingerprinting. Nucleic Acids Research 23:4407-4414.

Webster, M. S., and P. P. Marra. 2005. The importance of understanding migratory connectivity and seasonal interactions. The Johns Hopkins University Press.

Webster, M. S., P. P. Marra, S. M. Haig, S. Bensch, and R. T. Holmes. 2002. Links between worlds: unraveling migratory connectivity. Trends in Ecology & Evolution 17:76-83.

Weir, B. S., and C. C. Cockerham. 1984. Estimating F-Statistics for the Analysis of Population Structure. Evolution 38:1358-1370.

Wenink, P. W., A. J. Baker, and M. G. J. Tilanus. 1994. Mitochondrial Control- Region Sequences in 2 Shorebird Species, the Turnstone and the Dunlin, and their Utility in Population Genetic-Studies. Molecular Biology and Evolution 11:22-31.

Wennerberg, L. 2001. Genetic Variation and Migration of Waders. Ph.D., Lund University, Sweden, Lund.

Wennerberg, L., N. M. A. Holmgren, P.-E. Jönsson, and T. Von Schantz. 1999. Genetic and morphological variation in Dunlin Calidris alpina breeding in the Palearctic tundra. Ibis 141:391-398.

Wennerberg, L., M. Klaassen, and A. Lindstrom. 2002. Geographical variation and population structure in the White-rumped Sandpiper Calidris fuscicollis as shown by morphology, mitochondrial DNA and carbon isotope ratios. Oecologia 131:380-390.

Whitlock, M. C., and D. E. McCauley. 1999. Indirect measures of gene flow and

migration: FST≠1/(4Nm + 1). Heredity 82:117-125.

Winker, K. 2010. On the Origin of Species through Heteropatric Differentiation: A Review and a Model of Speciation in Migratory Animals. Ornithological Monographs No. 69:1-30.

Zink, R. M., A. Pavlova, S. Drovetski, and S. Rohwer. 2008. Mitochondrial phylogeographies of five widespread Eurasian bird species. Journal of Ornithology 149:399-413.

2.8. Tables

Barrow Siberia Kotzebue Nome YK- California Western Yucatan Puerto Panama Ecuador mtDNA n=120 Delta Mexico Rico adult 7 9 - 7 10 - - - - - 3 Females juvenile - - - - - 4 - - - 10 4 adult 9 3 - 12 8 9 - - 4 8 - Males juvenile - - - - - 11 - - 2 - -

68 TOTAL 16 12 - 19 18 24 - - 6 18 7

AFLP n=169 adult 9 10 6 9 11 - 10 8 - 10 3 Females juvenile - - - - - 4 - 9 - 84 adult 9 6 5 11 8 6 - - - 4 - Males juvenile - - - - - 6 7 6 -- - TOTAL 18 16 11 20 19 16 17 23 - 22 7

Morphometrics n=878 (159) Females 18 19 9 125 120 9 21 70 - 59 6 (9) (10) (6) (7) (9) (4) (10) (17) - (18) (6) Males 40 20 5 136 122 12 43 31 - 13 - (9) (5) (5) (8) (8) (12) (7) (6) - (3) - TOTAL 58 39 14 261 242 21 64 101 - 72 6 (18) (15) (11) (15) (17) (16) (17) (23) - (21) (6)

Table 2-2. Occurrence of each haplotype. Each row lists the total number of individuals with the respective haplotype in the sample as a whole and across the different sampling sites. For non-breeding sites, numbers are listed separately for each sex. The right-most column indicates corresponding haplotypes found by Enríquez-Paredes et al. (2012).

corresponding Puerto Panama haplotype Haplo- TOTAL YK- California Ecuador Barrow Nome Siberia Rico Enríquez- type Number Delta Female Male Female Male Female Male Paredes et al. 2012 study Hap01 55 8 6 6 9 2 8 2 6 4 4 WeSa001 Hap02 7 1 3 0 1 0 0 0 0 1 1 WeSa002 Hap03 5 0 2 1 0 0 1 1 0 0 0 Hap04 4 0 0 1 0 0 1 2 0 0 0 WeSa007 Hap05 3 0 0 0 0 0 2 0 0 1 0 WeSa027 Hap06 2 1 0 0 0 0 0 0 0 1 0 Hap07 2 1 0 0 0 0 1 0 0 0 0 Hap08 2 1 0 0 0 0 1 0 0 0 0 WeSa006 Hap09 2 1 0 0 0 0 0 0 1 0 0 Hap10 2 1 0 0 1 0 0 0 0 0 0 Hap11 2 0 1 0 0 0 1 0 0 0 0 Hap12 2 0 1 0 1 0 0 0 0 0 0 WeSa010 Hap13 2 0 1 0 1 0 0 0 0 0 0 Hap14 2 0 1 0 0 0 0 0 0 1 0 Hap15 2 0 1 0 1 0 0 0 0 0 0 WeSa003 Hap16 2 0 0 0 2 0 0 0 0 0 0 Hap17 1 1 0 0 0 0 0 0 0 0 0 Hap18 1 1 0 0 0 0 0 0 0 0 0 Hap19 1 0 1 0 0 0 0 0 0 0 0 Hap20 1 0 1 0 0 0 0 0 0 0 0 Hap21 1 0 1 0 0 0 0 0 0 0 0 Hap22 1 0 0 1 0 0 0 0 0 0 0 Hap23 1 0 0 1 0 0 0 0 0 0 0 Hap24 1 0 0 1 0 0 0 0 0 0 0

corresponding Puerto Panama haplotype Haplo- TOTAL YK- California Ecuador Barrow Nome Siberia Rico Enríquez- type Number Delta Female Male Female Male Female Male Paredes et al. 2012 study Hap25 1 0 0 1 0 0 0 0 0 0 0 Hap26 1 0 0 0 1 0 0 0 0 0 0 Hap27 1 0 0 0 1 0 0 0 0 0 0 Hap28 1 0 0 0 0 0 0 0 0 0 1 Hap29 1 0 0 0 0 0 0 0 0 0 1 Hap30 1 0 0 0 0 0 0 0 1 0 0 Hap31 1 0 0 0 0 0 0 0 1 0 0 Hap32 1 0 0 0 0 0 0 0 1 0 0 Hap33 1 0 0 0 0 1 0 0 0 0 0 Hap34 1 0 0 0 0 1 0 0 0 0 0 WeSa008 Hap35 1 0 0 0 0 0 0 1 0 0 0 70 Hap36 1 0 0 0 0 0 1 0 0 0 0 Hap37 1 0 0 0 0 0 1 0 0 0 0 Hap38 1 0 0 0 0 0 1 0 0 0 0 WeSa033 Hap39 1 0 0 0 0 0 1 0 0 0 0 Hap40 1 0 0 0 0 0 1 0 0 0 0 TOTAL 120 16 19 12 18 4 20 6 10 8 7

FST FST FST estimate based allele frequency DNA type analogue value 95% CI P on estimate Program Hickory (f- nuclear θ(II) 0.047 0.027-0.073 - allele frequency Bayesian free model) nuclear θ 0.052 0.036-0.069 - allele frequency Taylor expansion* TFPGA

band frequency nuclear F 0.027 - <0.001 - Arlequin ST (Dice)

mitochondrial FST 0.2 <0.001 frequency - Arlequin 71 Breeding

mitochondrial ΦST -0.009 0.781 molecular distance - Arlequin frequency - mitochondrial F 0.3 <0.001 - Arlequin ST singletons removed Hickory (f- nuclear θ(II) 0.032 0.012-0.051 - allele frequency Bayesian free model) nuclear θ 0.145 0.115-0.175 - allele frequency Taylor expansion* TFPGA band frequency nuclear F 0.016 - <0.05 - Arlequin ST (Dice) breeding ‐ mitochondrial FST 0.213 <0.001 frequency - Arlequin

Non mitochondrial ΦST -0.009 0.643 molecular distance - Arlequin frequency - mitochondrial F 0.356 <0.001 - Arlequin ST singletons removed

* Lynch & Milligan 1994

Barrow Nome Siberia YK-Delta California Puerto Rico Panama Ecuador Female Male Male Female Male Female Barrow . 0.231 0.404 0.308 0.315 0.274 0.227 0.468 0.267 0.410 Nome 0.171 . 0.317 0.233 0.212 0.200 0.135 0.377 0.183 0.317 Siberia 0.230 0.165 . 0.401 0.473 0.363 0.350 0.620 0.375 0.562 YK-Delta 0.238 0.176 0.235 . 0.315 0.275 0.230 0.462 0.268 0.407 Female 0.210 0.131 0.205 0.216 . 0.268 0.194 0.597 0.255 0.499 California

72 Male 0.192 0.132 0.187 0.197 0.158 . 0.187 0.424 0.230 0.366 Puerto Rico Male 0.164 0.092 0.158 0.171 0.111 0.117 . 0.445 0.164 0.350 Female 0.278 0.210 0.278 0.282 0.271 0.232 0.216 . 0.453 0.666 Panama Male 0.225 0.156 0.221 0.230 0.195 0.179 0.145 0.277 . 0.377 Ecuador Female 0.255 0.184 0.253 0.260 0.236 0.207 0.180 0.312 0.249 .

MseI primer EcoRI primer Number of ending in: ending in: polymorphic bands Breeding Non-breeding CAG AGC 26 X X CAG ACA 17 X X CAC AGG 14 X X CAC ACC 12 X X CAT AGC 14 X - CAT ACA 12 X - CATG ACT 7 X X

73 CATG AAC 4 X X

Table 2-6. Population pair-wise comparisons of AFLP band patterns. Values above the diagonal represent Cavalli-Sforza and Edwards chord distances (Takezaki formula). These distances were employed as a measure of genetic distance in isolation by distance analyses and calculated within, but not between breeding and non-breeding ranges. Pairwise FST values are given below the diagonal and were derived from an analysis of band frequency based distance measure (Dice distances). As males and females may differ in migratory strategies, they were entered separately into the analysis for all non-breeding sites. FST values that are significant at P, q and lFDR < 0.05 are given in bold.

Barrow Kotzebue Nome Siberia YK- California West. Mexico Yucatan Panama Ecuador Delta Male Female Male Female Male Female Male Female Female Barrow . 0.035 0.040 0.052 0.054 ------Kotzebue -0.006 . 0.033 0.048 0.050 ------

74 Nome 0.012 -0.010 . 0.043 0.046 ------Siberia 0.042 0.025 0.024 . 0.044 ------YK-Delta 0.060 0.043 0.039 0.041 . ------Male 0.019 0.018 0.023 0.076 0.073 . . California 0.037 0.035 0.045 0.047 0.043 Female 0.065 0.042 0.036 0.049 0.062 0.084 . Western Male 0.019 0.023 0.018 0.052 0.064 0.007 0.087 . 0.027 0.034 0.032 0.028 Mexico Female 0.018 0.023 0.020 0.066 0.025 -0.002 0.058 -0.020 . 0.037 0.036 0.035 Male 0.032 0.062 0.038 0.055 0.054 0.035 0.067 -0.017 0.016 . . Yucatan 0.035 0.032 Female 0.044 0.025 0.032 0.056 0.058 0.023 0.057 -0.015 -0.002 0.029 . Male 0.035 0.088 0.059 0.049 0.023 0.100 0.095 0.065 0.082 0.043 0.037 . . Panama 0.033 Female 0.045 0.019 0.035 0.073 0.031 0.025 0.053 -0.009 -0.021 0.005 -0.004 0.046 . Ecuador Female 0.035 0.041 0.024 0.024 0.014 0.059 0.061 -0.001 0.010 0.019 -0.023 -0.021 -0.001 .

2.9. Figures

Figure 2-2. Median-joining haplotype network. Haplotypes are denoted by circles, each connecting line corresponds to one mutation. Circle size is proportional to the number of individuals with this haplotype and colors indicate the subsample(s) in which each haplotype was found. Small black circles designate presumed intermediate haplotypes not present in the sampled individuals. The central, most common haplotype (Hap01) was found across all sites and was present in 45.8% of individuals. The remaining 39 haplotypes were relatively rare and differed in only 1-4 nucleotides from Hap01, resulting in a star-shaped network pattern that may indicate recent demographic expansion.

Figure 2-3. Distribution of haplotypes across breeding sites. The four circles show the haplotype composition at each of the four breeding sites. Circle sizes are proportionate to sample sizes. Each circle segment denotes a different haplotype with segment size indicating the proportion of individuals that possess the respective haplotype. Haplotypes that occurred at more than one site (including non-breeding sites) are shown in colour. White segments signify haplotypes that were found exclusively at the respective site. Hap01 was present at all sites, with a frequency of 32% among individuals at the Nome site and reaching 50% at the other three sites. Hap02 was shared among three, Hap03, 10, 12, 13 and 15 among two breeding sites. Hap04, 06, 07, 08, 09, 11 and 14 were found at only one breeding site, but were also present in individuals sampled on the non- breeding grounds (Fig. 2-4).

0.060 0.050

0.055 0.045

0.050 0.040

0.045 0.035 Genetic Distance 0.040 Genetic Distance 0.030

0.035 0.025

0.030 0.020 200 400 600 800 1000 1200 1400 0 2000 4000 6000 Geographic Distance Geographic Distance . a. Breeding range b. Non-breeding range

Figure 2-5.Correlation between genetic and geographic distances. Mantel tests confirmed a significant positive correlation between genetic and geographic distance among both the breeding (a) and the non-breeding (b) sites (Z=299, r=0.582, P=0.032, and Z=1721, r=0.620, P=0.022, respectively). Geographic distance accounted for 34% of the variation in genetic distance among breeding sites, strongly suggesting an isolation-by-distance pattern. This pattern was mirrored across the non-breeding sites where geographic distance accounted for 39% of the variation in genetic distance.

c Figure 2-6. Geographical representation of the first three spatial Principal Component scores on the breeding grounds. Individual spatial Principal Component (sPC) scores, derived from AFLP profiles, are shown on a map to visualize spatial relationships. The analysis is based on a modified Principal Component Analysis that explicitly takes spatial information into account. Individuals sampled at five breeding sites are represented by squares. Square color and size reflect individual sPC scores: white squares correspond to negative, black squares to positive values and size indicates magnitude. Consequently, a large white square vs. a large black square represents a greater genetic dissimilarity than a large vs. a small white square or a small white vs. a small black square. To achieve a better graphical representation, points were randomly scattered around the respective sampling sites and hence do not reflect exact sampling locations. The inset eigenvalues barplot depicts the magnitude of the eigenvalues of each score – positive values for global scores and negative for local scores. The respective eigenvalues for the scores depicted in each map are marked black. a. Increasing sPC1 scores from south to north indicate a latitudinal cline across the Alaskan breeding sites. The pattern is suggestive of isolation by distance. b. sPC2, on the other hand, appears to highlight differences between the easternmost breeding site in Siberia and the Alaskan breeding sites, with the geographically closer sites (Kotzebue and Nome) appearing less dissimilar than the more distant ones (YK-Delta and Kotzebue). c. The last retained global score (sPC3) appears to differentiate the centrally located Nome and, to a lesser degree, the Kotzebue site from the peripheral breeding sites.

Breeding range:

Non-breeding range:

Females Males Figure 2-8. Correlation between individual body size and genetic scores for females and males. The morphometric PC1 is based on culmen, tarsus and wing length, reflecting body size. Genetic scores are derived from a spatial PCA, reflecting the sPC1s for breeding and non-breeding samples respectively. For breeding females no relationship between body size and genetic scores was detected (r=0.036, P=0.411, n=41), whereas a moderately strong, negative correlation was present among males (r=-0.356, P=0.018, n=35). Across the non-breeding range, both sexes showed significant, inverse correlations among body size and genetic scores (r=-0.280, P=0.019, n=55 and r=-0.344, P=0.036, n=28).

2 r2=0.422 r =0.528

2.10. Appendix

Figure 2-10. Eigenvalues plots for breeding (a.) and non-breeding (b.) range spatial Principal Component Analyses. The screeplots depict the decomposition of the eigenvalues into their variance and autocorrelation components. The vertical, dashed line on the right represents the largest achievable variance through a linear combination of AFLP markers (= that of the first PC if a regular PCA is conducted on these markers), whereas the dashed horizontal lines at the top and bottom depict the respective limits for Moran’s I. For the breeding sites sPCA, the first three components contain the highest degree of variability and spatial structuring, clearly stood out from the remaining eigenvalues and were therefore retained after global analysis indicated significant global structure. For the non-breeding sites sPCA only the first component contained sufficient spatial and variance information to be interpreted after a significant global test.

Figure 2-11. AFLP sPC scores for breeding individuals. The first sPC is shown in relation to the second (left graph) and third (right graph) sPCs. While individuals cluster by site, a substantial degree of overlap exists, particularly among neighbouring sites. SPC1 scores show a gradual increase from the southernmost (YK-Delta) to the northernmost (Barrow) sampling site, indicating a cline across the Alaskan breeding sites. SPC2 and 3 accentuate the easternmost (Siberia) and central (Nome) breeding sites respectively.

Figure 2-12. AFLP sPC1 scores for non-breeding individuals in relation to latitude for each sex and age group. SPC1 scores show a consistent increase with latitude across all of the four sex and age groups.

Table 2-7. Haplotype definitions. Only positions that were variable in our sample are listed, with position 1 corresponding to the first nucleotide of the sequence analyzed in this study. The last column lists corresponding haplotypes found by Enríquez- Paredes et al. (2012).

Mutated Position 1 1 1 1 1 1 1 1 1 2 3 4 5 5 5 6 6 6 6 6 6 6 corresponding 1 1 2 2 2 2 3 3 3 4 6 7 8 9 1 1 6 6 7 7 8 8 9 6 6 9 2 7 9 2 2 6 6 6 6 7 haplotype Enríquez- 0 3 0 3 7 9 5 8 9 0 9 2 8 1 0 5 0 6 1 7 0 6 2 8 4 5 7 1 7 8 9 0 4 5 7 4 Paredes et al. 2012 Hap01 C C C C C C A A A TTTCCACCGCGATGGA T C TTTTGTCTG WeSa001 Hap02 ...... C ...... WeSa002 Hap03 ...... C . . . . . Hap04 T ...... WeSa007 Hap05 ...... C ...... T . . WeSa027 Hap06 ...... C ...... Hap07 T . . . T ...... A Hap08 ...... G ...... WeSa006 88 Hap09 ...... A . . . . Hap10 ...... T . . Hap11 ...... A ...... Hap12 ...... T ...... WeSa010 Hap13 . . T ...... G . A ...... Hap14 ...... C . . . . Hap15 ...... C ...... WeSa003 Hap16 . . . . . T ...... Hap17 ...... A ...... T ...... Hap18 ...... G . T ...... Hap19 T . . . . T . . . . . C ...... A Hap20 . T ...... C T . . Hap21 . . . T . . . G ...... A ...... Hap22 . . . . . T . . . . . C ...... Hap23 ...... G ...... Hap24 ...... G . T ...... G .

Mutated Position 1 1 1 1 1 1 1 1 1 2 3 4 5 5 5 6 6 6 6 6 6 6 corresponding 1 1 2 2 2 2 3 3 3 4 6 7 8 9 1 1 6 6 7 7 8 8 9 6 6 9 2 7 9 2 2 6 6 6 6 7 haplotype Enríquez- 0 3 0 3 7 9 5 8 9 0 9 2 8 1 0 5 0 6 1 7 0 6 2 8 4 5 7 1 7 8 9 0 4 5 7 4 Paredes et al. 2012 Hap25 . T ...... C ...... Hap26 ...... G . Hap27 ...... T ...... Hap28 ...... T . . . A ...... Hap29 ...... G ...... Hap30 ...... T ...... Hap31 ...... G ...... Hap32 ...... G ...... Hap33 . . . . T ...... Hap34 ...... C . . T ...... T . . WeSa008 Hap35 ...... T . T ...... Hap36 . . T ...... C ...... 89 Hap37 ...... T ...... Hap38 ...... C ...... WeSa033 Hap39 ...... A ...... A ...... Hap40 ...... C ......

Table 2-8. Relative frequency of occurrence of each haplotype. The overall frequency within the sample and relative occurrence within each site are given in percent. For non-breeding sites, percentages are listed separately for each sex. The right- most column indicates corresponding haplotypes found by Enríquez-Paredes et al. (2012).

YK- California Puerto Rico Panama Ecuador corresponding TOTAL Barrow Siberia Nome haplotype Enríquez- Haplotype Delta Female Male Male Female Male Female percentage (%) (%) (%) Paredes et al. 2012 (%) (%) (%) (%) (%) (%) (%) Hap01 46 50 50 32 50 50 40 33 60 50 57 WeSa001 Hap02 6 6 0 16 6 0 0 0 0 13 14 WeSa002 Hap03 4 0 8 11 0 0 5 17 0 0 0 Hap04 3 0 8 0 0 0 5 33 0 0 0 WeSa007 Hap05 3 0 0 0 0 0 10 0 0 13 0 WeSa027 Hap06 2 6 0 0 0 0 0 0 0 13 0 Hap07 2 6 0 0 0 0 5 0 0 0 0 Hap08 2 6 0 0 0 0 5 0 0 0 0 WeSa006 Hap09 2 6 0 0 0 0 0 0 10 0 0 90 Hap10 2 6 0 0 6 0 0 0 0 0 0 Hap11 2 0 0 5 0 0 5 0 0 0 0 Hap12 2 0 0 5 6 0 0 0 0 0 0 WeSa010 Hap13 2 0 0 5 6 0 0 0 0 0 0 Hap14 2 0 0 5 0 0 0 0 0 13 0 Hap15 2 0 0 5 6 0 0 0 0 0 0 WeSa003 Hap16 2 0 0 0 11 0 0 0 0 0 0 Hap17 1 6 0 0 0 0 0 0 0 0 0 Hap18 1 6 0 0 0 0 0 0 0 0 0 Hap19 1 0 0 5 0 0 0 0 0 0 0 Hap20 1 0 0 5 0 0 0 0 0 0 0 Hap21 1 0 0 5 0 0 0 0 0 0 0 Hap22 1 0 8 0 0 0 0 0 0 0 0 Hap23 1 0 8 0 0 0 0 0 0 0 0 Hap24 1 0 8 0 0 0 0 0 0 0 0 Hap25 1 0 8 0 0 0 0 0 0 0 0

Table 2-9. AFLP primer pairs run in pooled approach. Each of the listed primer combinations marked with an X were surveyed for fixed genetic differences among juvenile females sampled at northern, southern and eastern non-breeding sites. No fixed differences were detected for any of the thousands of possible markers. Marker combinations denoted by an asterisk were run in the single-lane approach, a hyphen indicates un-tested combinations.

EcoRI primer ending in:

AGC ACA AGG ACC ACT AAC CAG * * X X - - CAT * * X X X X CAC X X * * X X CAA X X X X X X CATG X X X X * * CATC X X X X X X

92 CACT - - X X X X CAGC X X - - - - CACG X X X X X X CATT X X X X X X CAAC - - X X X X I primer ending in: CAAG X X X X X X Mse CACC X X X X X X CACA X X X X X X CAAA X X X X X X CAAT X X X X X X CAGA X X - - X X

Table 2-10. Component loadings for Morphology PC1. All morphometric measurements loaded strongly and positively onto the first PC, suggesting that this component reflects overall size.

PC1 PC1

(individuals (all with genetic data) individuals)

Culmen 0.907 0.895

Tarsus 0.893 0.865

Wing 0.778 0.780 93

Chapter 3. Songs differ but alarm calls do not: a study of geographical variation in Western Sandpipers

3.1. Abstract

Being subject to sexual selection, mating signals such as song may not only show faster rates of intraspecific divergence than other traits, but may also play a crucial role in limiting gene flow among populations. In species with learned songs, the relationship between song and genetic population structure can be complex, i.e. song dialects are not necessarily indicative of genetic divergence nor its geographic pattern. A greater level of congruence is, however, expected for species with innate songs. Many shorebird species possess songs that appear to fulfill similar functions as oscine song but do not appear to be learned. Among these is the Western Sandpiper, a small, migratory and therefore highly mobile species that shows significant genetic population structure across its geographically limited breeding range. Given the species mobility, restricted breeding range and significant population structure, song may be expected to play a role in limiting gene flow across the breeding range. We therefore recorded male songs at four breeding sites at different latitudes to investigate geographic variation in song, contrasting it with an acoustic signal not expected to experience sexual selection (alarm calls). We conducted spectrographic analyses, measuring temporal, center frequency and bandwidth variables which we entered into Principal Component Analyses. Retained Principal Components were tested for differences among sites using ANOVA. Song characteristics varied geographically (ANOVA of PC1: P < 0.001) while alarm call characteristics did not. Songs appeared to increase with frequency and decrease in length from the southern

to northern breeding sites. The geographical variation in male songs suggested a latitudinal trend with gradual changes in certain song characteristics – a pattern that shows similarities with the main pattern found in the population genetic data that was based on AFLP markers (Chapter 2). This suggests that the observed differences in song may be associated with genetic divergence.

3.2. Introduction

A peacock’s tail, the song of a cicada, bird, monkey or whale, the flashing light of a firefly or the pheromone cocktail of a moth - these are but a few striking examples of the many elaborate, species-specific signals that play key roles in mate attraction and communication between potential mates and/or rivals. Their function in this context makes these signals prime candidates for sexual selection (West-Eberhard 1983; Ryan and Kime 2003). Sexually selected traits may evolve more rapidly than non-sexually selected traits and can strongly affect gene flow, leading to the establishment or maintenance of reproductive barriers (West-Eberhard 1983; Price 1998; Gonzalez-Voyer and Kolm 2011; Seddon et al. 2013). As such they can be expected to be among the first traits to show geographical differences and at the same time may constitute key elements in promoting further differentiation and ultimately facilitating speciation (West-Eberhard 1983; Panhuis et al. 2001).

Geographical variation of oscine bird song has been studied extensively (Price 2007; Catchpole and Slater 2008). As song learning is wide-spread among songbirds, however, the relationship between song dialects and genetic divergence can be complex and the extent to which geographic differences in song correlate with genetic differentiation varies widely among (and possibly even within) species e.g.(e.g. Baker 1975; Balaban 1988; Payne and Westneat 1988; Lougheed and Handford 1992; Lougheed et al. 1993; MacDougall-Shackleton and MacDougall-Shackleton 2001; Soha et al. 2004; Irwin et al. 2008). In fact, song learning may promote genetic divergence

under some conditions and hinder it under others (Slabbekoorn and Smith 2002; Price 2007). In contrast, a more direct relationship between genetic divergence and song can be expected in taxa that do not exhibit song learning (Miller 1996; Price 2007). In these taxa, geographical differences in song are more likely to indicate incipient or ongoing divergence. They may provide opportunities to study vocal communication in an evolutionary context without the added complexity that song learning entails. For instance, when we compare the geographic variability or rate of evolution of song as a sexually selected signal with that of other vocalization types, the results can be more readily interpreted in the absence of song learning.

Among the avian taxa that are thought to not learn their songs are shorebirds (von Frisch 1956; von Frisch 1959; Miller et al. 1983; Miller 1986; Douglas 1996; Miller 1996). Many shorebird species are highly vocal, and possess a wide repertoire of different vocalization types with a multitude of communicative functions and varying levels of complexity (Miller 1984). Some of these vocalizations (e.g. songs and courtship vocalizations) likely evolve under the influence of sexual selection, while others, such as alarm calls, should not. The vocal repertoire of our study species, the socially monogamous Western Sandpiper (Calidris mauri), encompasses both types of vocalizations. Song can be heard on the breeding grounds in Western Alaska and on the Chukchi Peninsula of eastern Russia where it is often given by males performing display flights over their territories making it an ideal candidate for detecting geographic variation (Brown 1962; Holmes 1973; Miller 1996). It consists of a series of introductory notes that are followed by a descending trill (Brown 1962; Holmes 1973). While it is frequently associated with display flights, it is also given in a variety of other contexts e.g. while courting a female on the ground and during aerial chases and appears to function in mate attraction and territorial defense (Brown 1962; Lanctot et al. 2000; Chapter 5). Both of these functions would be expected to closely link male song with reproductive success, thereby subjecting it to sexual selection pressures. In contrast, adult alarm calls are given by both sexes when potential predators approach the nest or young

(Holmes 1973; Johnson et al. 2008). We investigated whether male song differs geographically across the breeding range in Western Sandpipers and whether such variation was greater than that occurring in alarm calls, a presumably non-sexually selected trait.

Previous studies suggested the possible existence of distinct populations with different migration and life history strategies in this species, based on morpholometric, life history and genetic data from the non-breeding grounds (Haig et al. 1997; Fernández et al. 2004; Nebel 2005; O'Hara et al. 2005; Lank and Nebel 2006; O'Hara et al. 2006; Enríquez-Paredes et al. 2012). Stable isotope data, on the other hand, failed to show the strong migratory connectivity expected under this scenario (Franks et al. 2012). As shown in Chapter 2, we found that Western Sandpipers show low but significant population structure across the breeding and non-breeding grounds in both mitochondrial and genomic (AFLPs; Amplified Fragment Length Polymorphisms) markers. The structure revealed by the latter, appears to reflect primarily an isolation by distance pattern (Chapter 2). The mitochondrial DNA control sequence data, on the other hand, suggested haplotype frequency variations across sites, but showed no clear geographic pattern. Given the species’ high mobility and limited breeding range in Western Alaska and far eastern Siberia (Wilson 1994), patterns of song variation might not only provide further evidence for or against differentiation, but could – if such differentiation exists – be a crucial mechanism in maintaining reproductive barriers.

In light of the short evolutionary time-frames suggested by the genetic data (Chapter 2) and without an a priori expectation of directional selection, we do not expect alarm calls to have diverged significantly among different breeding populations. Given the postulated close association of a sexually selected trait such as song with population genetic structure, we would expect its geographic patterns to more closely reflect those of the genetic data. We therefore hypothesize that a.) male Western Sandpipers will show

greater geographic variation in song than in alarm calls and b.) that male song will exhibit gradual spatial variation, consistent with an isolation by distance pattern.

3.3. Methods

We recorded male songs and alarm calls at four breeding sites in Alaska: the Yukon-Kuskokwim-Delta (YK-Delta; 61°21'N, 165°07'W; June 11-15, 2009), near Nome (64°26'N, 164°56'W; May 21-26, June 07, 2009 and June 14-15, 2010 (alarm calls only, two males)), Wales (65°36'N, 168°05'W; May 29 – June 04) and Kotzebue (66°50'N, 162°34'W; June 11-18). At the Nome, Wales and Kotzebue sites, we used a Marantz PMD 670 solid state recorder with a Sennheiser ME62 microphone that was mounted inside a parabolic dish (56cm diameter) to obtain the recordings. A different microphone of the same make and model was used at the YK-Delta site in conjunction with a parabolic dish of the same size (56cm diameter) and a Marantz PMD 660 solid state recorder (see Appendix for a discussion of the potential effects of using two different microphones). We recorded at a sampling rate of 48kHz and a bit depth of 16 and files were saved in WAVE (.wav) format. Recording distances were estimated to the nearest 5m and ranged from 5 to 35m with one exception where a male may have been recorded at a range of up to 105m. To reduce the likelihood of repeatedly recording the same individual we attempted to keep a minimum distance of 70m between any two recordings of unmarked males, unless these males could clearly be distinguished by direct observation or by their association with different nests or chicks of distinctly different ages. To ensure that recording distances did not affect our results we conducted Kruskal-Wallis-tests on median recording distances; these did not differ between sites for songs (H = 4.03, df = 3, P = 0.258) nor for alarm calls (H = 2.59, df = 2, P = 0.274). For alarm calls we could only compare recording distances for three sites due to missing distance data for the YK-Delta. We identified males based on their behaviour or, in the

case of individually marked (color bands or flags) birds, based on culmen measurements or genetic sex.

3.3.1. Spectrographic Analysis:

We viewed spectrograms in Raven (1.4; Bioacoustics Research Program 2011) and gave each recording a quality score on a scale from 1-5 based on amplitude, signal- to-noise ratio, and interference of other signals. Solely good quality recordings with a score of 3 or higher were used for subsequent analyses in Raven 1.4 with the following settings: Hann window, window size of 312 points with 90% overlap, grid spacing 46.9Hz.

3.3.2. Song:

We analyzed a total of 413 songs from 45 males (YK-Delta: n = 12, Nome: n = 14, Wales: n = 11 and Kotzebue: n = 8 males). For our quantitative analysis of song characteristics, we focused solely on the fundamental frequency in the frequency domain (ignoring harmonics). For each song we determined total duration, duration of the introductory and trill parts and 90% bandwidth of the first (bandwidth 1) and last (bandwidth 2) part of the song. Additionally, we measured frequency and temporal characteristics of specific song components that were selected based on their presence in all or nearly all of the recorded songs (Fig. 3-1). For a detailed description and complete list of variables see Appendix Table 3-3. Purely for descriptive purposes we also calculated the overall center frequency (the mean of all center frequency measurements), trill rate during the first 10 and last 12 pulses of the trill and maximum bandwidth (the greater of bandwidth 1 and 2).

3.3.3. Alarm calls:

A total of 193 Alarm call recordings from 25 males (YK-Delta: n = 6, Nome: n = 7, Wales: n = 5 and Kotzebue: n = 7 males) were of sufficient quality to be analyzed. Again, we restricted our analysis to the fundamental frequency. We measured the duration, peak and center frequency and 90% bandwidth of the first, third and last notes and the entire call, as well as the total number of notes (Fig. 3-2; Appendix Table 3-3)

3.3.4. Statistical tests

We conducted separate Principal Component Analyses on the quantitative song and alarm call characteristics. Only the mean values for each male were entered into these analyses. As parallel analysis has been shown to be a more reliable method of determining the number of factors to retain than either Catell’s scree plot or Kaiser’s eigenvalue>1 methods, we ran a parallel analysis using the ViSta-PARAN plug-in in ViSta (Young 2003; Ledesma and Valero-Mora 2007). Based on the results from this analysis we retained the first three Principal Components (PCs) for both the song and alarm call analysis and conducted ANOVAs with post-hoc Tukey's HSD tests on these PCs to test for differences between sites. PCs that differed significantly between sites were also subjected to a variance component analysis to determine the proportion of variation allocatable to this geographical difference (the inter-site component) versus the inter- individual component of variation within each site. To allow us to also estimate the intra- individual component of variation, we used the factor loadings from the PCA on individual male means to calculate PC scores for each song. To make this calculation possible missing data values (0.5 % of the data) were replaced with the respective variable means for the individual. Finally, we obtained a measure of the degree of variability of the two vocalization types by calculating coefficients of variation (COVs) for the original variables. We then calculated mean COVs for each of the following variable types: peak frequency, center frequency, bandwidth and duration.

100

3.4. Results

3.4.1. Song

The songs of male Western Sandpipers showed the same basic structure across the four different breeding sites (Fig. 3-3): they were composed of a variable number of introductory notes that culminated in a trill. Despite this structural conformity, quantitative differences in song characteristics existed across the breeding range. Basic song characteristics for each site are shown in Table 3-1, the means for all variables included in the PCA can be found in the appendix (Appendix Table 3-3). The first three components of the PCA were retained and together accounted for 63% of the total variation. PC2 and PC3 which accounted for 12.7 and 10.3% respectively did not differ among sites. PC2 primarily represented bandwidth variables (positive loadings). Other bandwidth measures loaded positively onto PC3 along with the number of introductory notes (positive) and their duration (negative). All center and peak frequency variables, loaded positively onto PC1, in addition to song and trill duration which had negative loadings (Appendix Table 3-4). PC1 encompassed a considerably larger proportion of the total variation (40%) and PC1 scores differed significantly between sites (F = 15.63, df = 3 and 41, P < 0.001; Tukey post- hoc test significant at P<0.001, <0.01 and <0.001 for YK-Delta vs. Nome, Wales and Kotzebue and P<0.05 for Wales vs. Kotzebue; Fig. 3-4). Specifically, male songs at the southernmost site in the YK-Delta differed from those at all other sites and songs recorded near Wales differed from those at the Kotzebue site (Fig. 3-4).

In accordance with these results, the variance component analysis of PC1 attributed a substantial part of the total variation to the between-sites component (46.3%), while the between-males and within-individual components were responsible for 30.0 and 23.8% respectively.

101

3.4.2. Alarm calls

Alarm calls consisted of a variable number of notes with a rich harmonic structure and did not differ significantly between the four breeding sites (Figs.3-3 and 3-5). Mean values for duration, number of notes and center/peak frequency for each site are listed in Table 3-1, means for all variables in Appendix Table 3-3. The PCA conducted on these variables yielded three retainable factors accounting for 73.2% of the total variation. Of these, PC1 accounted for 43.4% of the variation, and again was driven by frequency variables (Appendix Table 3-4). Note duration loaded positively onto PC2 (15.9% of the total variation) and the main loadings for PC3 (13.9% of the total variation) included bandwidth and duration variables. In contrast to the song analysis, the PCA scatterplot indicates that alarm calls from all sites overlap in their characteristics and no separation among sites was evident. Accordingly, no differences between sites were found for any of the Principal Components (PC1: F = 0.23 , df = 3 and 21, P = 0.874; PC2: F = 0.56, df = 3 and 21, P = 0.646 ; PC3: F = 1.32, df = 3 and 21, P = 0.294). None of the alarm call mean COV values for each variable type (peak and center frequency, bandwidth, duration) were lower than the corresponding COV values for song (Table 3-2).

3.5. Discussion

Our spectrographic analysis clearly demonstrates that male song characteristics differ across the breeding range. Males at the southernmost site, in particular, sang songs that differed from those at all other sites (for a discussion of potential microphone effects see the Appendix). While the PCs derived from our multivariate analysis are composite variables and should be interpreted with caution, factor loadings can give us an indication of the nature of the respective PC (Stevens 2002; Field 2009; Afifi et al. 2012). Factor loadings indicate the degree of correlation between the PC and the underlying variables and can be used as a measure of the relative contribution of these variables towards the PC (Stevens 2002; Field 2009; Afifi et al. 2012). PC1, which differed significantly

102

between sites, was strongly driven by frequency characteristics (positive correlation) and total song and trill duration (negative correlation). The YK-Delta songs, which had significantly smaller PC1 scores, had the lowest mean frequencies and longest duration, while the songs at the northernmost site, Kotzebue, had the highest frequencies and shortest duration. As songs at the two central sites (Wales and Nome) showed intermediate values this might reflect a latitudinal trend. The mean PC1 for songs recorded at Nome, however, lay somewhat higher – though not significantly so – than that at Wales, even though Wales is located further north. Additional song recordings from breeding sites at the northern or southern edges of the breeding range could help to clarify whether there is indeed a relationship between latitude and song frequency or duration.

What might drive such a pattern is difficult to determine without additional data. Due to its functional role in mate attraction and territorial defense, Western Sandpiper song may be shaped not just by natural selection but also by sexual selection or by an interplay of both. The intensity of sexual selection, for instance, can vary latitudinally e.g. due to differing lengths in breeding season duration which may drive geographical variation in song features, as shorter time frames may translate to more intense sexual selection (Read and Weary 1992; Irwin 2000). As mate attraction in many shorebirds also involves aerial displays the potential effects of this variation on song could also be indirect (via morphology, see discussion below). Latitudinal changes in habitat, on the other hand, could give rise to differing natural selection pressures. Again, both indirect (via morphology) or direct influences (via sound transmission) on song are possible. The concept of signal space predicts that species occupy different niches in a multidimensional acoustic space – as they would for other ecological factors – and shape their signal characteristics to avoid overlap with other species and maximize transmission and recognisability (Nelson and Marler 1990; Ryan and Kime 2003; Luther 2008). Frequency characteristics appear to provide particularly important discrimination cues in this regard (Nelson 1988). If sites differ in their species composition and breeding

103

densities, this could result in a substantially different acoustic environment, altering the shape of the adaptive landscape and leading to differing optimal signal solutions (Fear and Price 1998). In addition to these biotic sounds, abiotic sources of background noise and habitat characteristics that influence sound transmission may play a decisive role in shaping acoustic signals (e.g. Morton 1975; Wiley and Richards 1978; Dubois and Martens 1984; Seddon 2005). Songbird studies have primarily dealt with the transmission differences between open habitats such as grasslands and closed habitat types, e.g. forests (e.g. Morton 1975; Sorjonen 1986a, b). Most shorebird species inhabit open habitat types and male songs are often given during aerial displays, so differences in vegetation are less likely to account for geographic differences in shorebird song. There are, however, other habitat-related features that have the potential to shape song differences that may affect shorebirds. In Willets (Tringa semipalmata), the eastern subspecies (T. s. semipalmatus) sings shorter and higher songs than the western subspecies (T. s. inornatus), a difference which appears to have evolved as an adaptation to the specific acoustic conditions found in the eastern subspecies’ coastal breeding habitat (Douglas 1996; Douglas and Conner 1999). Wind also influences sound transmission, so differing wind conditions are another possible selective factor (Wiley and Richards 1978). Systematic recordings and evaluations of background noise and wind conditions at the different breeding sites could help to determine whether any of the aforementioned biotic or abiotic factors play a role in shaping the observed geographical differences in Western Sandpiper song. While the differences might represent vocal adaptations to local acoustic conditions, it is also possible that they reflect morphological differentiation between sites (e.g. body or culmen size). Frequency, for instance, may be negatively correlated with body size, but this correlation does not always hold true for within-species comparisons (Bradbury and Vehrencamp 1998; Fletcher 2004; Patel et al. 2010). For at least one shorebird species, the Least Sandpiper (Calidris minutilla), such a correlation appears to exist across different breeding populations (Miller 1986). In Western Sandpipers, females are the larger bodied sex and sing at lower frequencies than males (Cartar 1984; Sandercock 1998; Stein et al. 2008; unpubl. data). Whether this truly reflects a negative

104

correlation, however, needs to be tested directly by comparing individual body size indices and vocalization frequencies. If this correlation holds, geographical variation in song frequencies could represent an indirect effect, resulting from natural or sexual selection that is acting on body size rather than directly on the vocalizations themselves.

In contrast to the clear differences in male song, we could not detect any differences in alarm calls across sites. While our sample size for alarm calls was smaller overall, the PC scatter plot showed no evidence of clustering, making us confident that this result would not change with a larger sample size. The selection pressures and constraints that act on alarm calls are conceivably very different from those acting on song. Song is largely a long-distance communication signal, whereas alarm calls may act at medium range e.g. when alerting the chicks in the immediate vicinity to the presence of a predator or distracting predators (Miller 1996; Johnson et al. 2008). As such, alarm calls might have to adapt to a lesser degree to changing acoustic environments to maintain their ability to reach the intended receiver (Ryan and Kime 2003). Finally, and perhaps most importantly, alarm calls, unlike songs, should not be subject to sexual selection. At the species level, the alarm calls of calidrid sandpipers tend to show greater similarities between species than songs and rhythmically repeated calls do, which suggests that different types of vocalizations evolve at different speeds (Miller 1996).

Songs might differentiate faster than alarm calls because their evolution is driven by sexual selection. Conversely, alarm calls might evolve at a slower pace due to ecological selection pressures. Alarm call-specific constraints might prevent deviations from optimal call characteristics and even lead to convergent evolution across species (Marler 1955). In this case we would expect alarm call characteristics to be less variable than those of song or other vocalization types. We found no evidence for this in Western Sandpipers as mean coefficients of variation were consistently higher in alarm call than in song features (Table 4.2). If the song differences reflect genetic changes then our study

105

lends further support to the hypothesis that sexually selected signals tend to evolve faster than those that do not experience sexual selection pressures (West-Eberhard 1983).

Sexually selected traits can also play a crucial role in restricting gene flow, thereby maintaining genetic differentiation (West-Eberhard 1983; Price 1998; Ryan and Kime 2003; Seddon et al. 2013). For this to be the case in Western Sandpipers, the observed song differences would have to, firstly, be closely associated with genetic divergence and, secondly, be sufficiently large to be biologically meaningful. We are not aware of any study that has tested directly to what degree geographical differences in song reflect genetic differentiation in shorebirds. Based on the low geographical variation in all species studied to date, it has been suggested that shorebirds do not learn their songs and song differences therefore likely do reflect genetic differentiation (Miller et al. 1983; Miller 1986, 1996). Nevertheless, we cannot completely exclude the possibility that a certain degree of song learning takes place in some shorebird species. Song development in shorebirds is clearly an important aspect that requires further study. Even without explicit song learning, phenotypic plasticity could allow males to adjust their songs to match prevalent environmental conditions. In songbirds, intra-individual variability exists and may be positively associated with variable environments (Medina and Francis 2012). It is notable, however, that the geographical variation in Western Sandpiper song suggested a latitudinal trend with gradual changes in certain song characteristics – a pattern that shows similarities with the main pattern found in the AFLP based genetic data. This suggests that the apparent differences in song may indeed be associated with genetic divergence.

The second requirement, the biological significance of the observed song differences, also needs to be determined. The differences we found were purely quantitative (i.e. no differences in song structure were apparent) and the degree of song differentiation between sites appears relatively small when compared to the substantial differences that can be found in many songbird dialects (Marler and Slabbekoorn 2004;

106

Price 2007). It is however, comparable to or larger than the degree of differentiation found in other shorebird species for comparable vocalizations (Miller et al. 1983; Miller 1986; Douglas 1996). In all of these studies the respective shorebird species possessed broader breeding ranges, the differentiation was found to occur over larger geographic distances and in two out of three cases it involved different subspecies. In the Short-billed Dowitcher (Limnodromus griseus), individuals of three different subspecies that breed in Manitoba, Quebec and British Columbia, respectively, showed quantitative differences in their songs (Miller et al. 1983). As in our study, the general song structure did not differ between sites. In Least Sandpipers (Calidris minutilla), songs recorded at breeding sites across the length of the Canadian arctic were very similar (Miller 1986). A second type of vocalization that is given by males in aerial displays, the rhythmically repeated call (RRC), also showed very little geographic variation in this species. Most of the variance for this vocalization type was due to differences between males within sites; only a single frequency variable differed significantly between sites with an absolute difference of less than 150Hz (Miller 1986). As mentioned above, the songs of the eastern and western subspecies’ of the Willet also differ in their frequency and temporal characteristics (Douglas 1996). Research on this species demonstrates that quantitative differences can be biologically meaningful in shorebirds. Playback studies indicated that individuals of the eastern subspecies were able to discriminate between their songs and those of the western subspecies and adjusted their response accordingly (Douglas 1998). The quantitative differences in song found in Western Sandpipers thus have the potential to be biologically meaningful and to contribute towards reproductive isolation.

Ultimately, playback experiments would have to show whether individuals discriminate between songs recorded at different sites. Ideally, these experiments should measure the response of female Western Sandpipers to gauge the effect song differences may have on mate choice.

107

3.6. References

Afifi, A. A., S. May, and V. A. Clark. 2012. Practical multivariate analysis, 5 ed. Taylor & Francis Group, Boca Raton.

Baker, M. C. 1975. Song Dialects and Genetic Differences in White-Crowned Sparrows (Zonotrichia leucophrys). Evolution 29:226-241.

Balaban, E. 1988. Cultural and Genetic Variation in Swamp Sparrows (Melospiza georgiana) .1. Song Variation, Genetic Variation, and their Relationship. Behaviour 105:250-291.

Bioacoustics Research Program. 2011. Raven Pro: Interactive Sound Analysis Software (Version 1.4) [Computer software]. Ithaca, NY: The Cornell Lab of Ornithology. Available from http://www.birds.cornell.edu/raven. .

Bradbury, J. W., and S. L. Vehrencamp. 1998. Principles of animal communication. Sinauer Associates, Sunderland, Mass.

Brown, R. G. B. 1962. The Aggressive and Distraction Behaviour of the Western Sandpiper Ereunetes mauri. Ibis 104:1-12.

Cartar, R. V. 1984. A morphometric comparison of Western and Semipalmated sandpipers. Wilson Bulletin 96:277-286.

Catchpole, C. K., and P. J. B. Slater. 2008. Bird song: biological themes and variations. Cambridge University Press, Cambridge, UK.

Douglas, H. D. I. 1996. Communication, evolution and ecology in the Willet (Catoptrophorus semipalmatus): its implications for shorebirds (Suborder Charadrii). Master's Thesis, Wake Forest University Winston-Salem, NC.

108

Douglas, H. D. I. 1998. Response of Eastern Willets (Catoptrophorus s. semipalmatus) to vocalizations of Eastern and Western (C. s. inornatus) willets. Auk 115:514-518.

Douglas, H. D. I., and W. E. Conner. 1999. Is there a sound reception window in coastal environments? Evidence from shorebird communication systems. Naturwissenschaften 86:228-230.

Dubois, A., and J. Martens. 1984. A Case of Possible Vocal Convergence between Frogs and a Bird in Himalayan Torrents. Journal Für Ornithologie 125:455-463.

Fear, K. K., and T. Price. 1998. The adaptive surface in ecology. Oikos 82:440- 448.

Fernández, G., P. D. O'Hara, and D. B. Lank. 2004. Tropical and subtropical Western Sandpipers (Calidris mauri) differ in life history strategies. Ornitologia Neotropical 15:385-394.

Field, A. P. 2009. Discovering statistics using SPSS: Introducing statistical methods., 3 ed. SAGE, London.

Fletcher, N. H. 2004. A simple frequency-scaling rule for animal communication. Journal of the Acoustical Society of America 115:2334-2338.

109

Warnock, R. C. Ydenberg, and D. B. Lank. 2012. Range-wide patterns of migratory connectivity in the Western Sandpiper Calidris mauri. Journal of Avian Biology 43:155- 167.

Haig, S. M., C. L. GrattoTrevor, T. D. Mullins, and M. A. Colwell. 1997. Population identification of western hemisphere shorebirds throughout the annual cycle. Molecular Ecology 6:413-427.

Holmes, R. T. 1973. Social Behavior of Breeding Western Sandpipers Calidris mauri. Ibis 115:107-123.

Irwin, D. E. 2000. Song variation in an avian ring species. Evolution 54:998- 1010.

Irwin, D. E., M. P. Thimgan, and J. H. Irwin. 2008. Call divergence is correlated with geographic and genetic distance in greenish warblers (Phylloscopus trochiloides): a strong role for stochasticity in signal evolution? Journal of Evolutionary Biology 21:435- 448.

Johnson, M., S. Aref, and J. R. Walters. 2008. Parent-offspring communication in the Western Sandpiper. Behavioral Ecology 19:489-501.

110

Lank, D. B., and S. Nebel. 2006. Cross-cutting research on a flyway scale – beyond monitoring. Pages 107-112 in Waterbirds around the world. (G. C. Boere, C. A. Galbraith, and D. A. Stroud, Eds.). The Stationery Office, Edinburgh, UK.

Ledesma, R. D., and P. Valero-Mora. 2007. Determining the Number of Factors to Retain in EFA: an easy-to-use computer program for carrying out Parallel Analysis. Practical Assessment, Research & Evaluation 12: Available online: http://pareonline.net/getvn.asp?v=12&n=12.

Lougheed, S. C., and P. Handford. 1992. Vocal Dialects and the Structure of Geographic Variation in Morphological and Allozymic Characters in the Rufous- Collared Sparrow, Zonotrichia capensis. Evolution 46:1443-1456.

Lougheed, S. C., P. Handford, and A. J. Baker. 1993. Mitochondrial-DNA Hyperdiversity and Vocal Dialects in a Subspecies Transition of the Rufous-Collared Sparrow. Condor 95:889-895.

Luther, D. A. 2008. The Evolution of Communication in a Complex Acoustic Environment. Ph.D. Thesis, The University of North Carolina, Chapel Hill.

MacDougall-Shackleton, E. A., and S. A. MacDougall-Shackleton. 2001. Cultural and genetic evolution in mountain white-crowned sparrows: Song dialects are associated with population structure. Evolution 55:2568-2575.

Marler, P. 1955. Characteristics of some Animal Calls. Nature 176:6-8.

Marler, P., and H. Slabbekoorn, Eds. 2004. Nature's music: the science of birdsong. Elsevier Academic, London.

Medina, I., and C. D. Francis. 2012. Environmental variability and acoustic signals: a multi-level approach in songbirds. Biology Letters 8:928-931.

111

Miller, E. H. 1984. Communication in breeding shorebirds. Pages 169-241 in Shorebirds: Breeding Behavior and Populations (J. Burger, Olla, B.L., Eds.). Plenum Press, New York.

Miller, E. H. 1986. Components of Variation in Nuptial Calls of the Least Sandpiper (Calidris minutilla; Aves, Scolopacidae). Systematic Zoology 35:400-413.

Miller, E. H., W. W. H. Gunn, and R. E. Harris. 1983. Geographic Variation in the Aerial Song of the Short-billed Dowitcher (Aves, Scolopacidae). Canadian Journal of Zoology-Revue Canadienne De Zoologie 61:2191-2198.

Morton, E. S. 1975. Ecological Sources of Selection on Avian Sounds. American Naturalist 109:17-34.

Nebel, S. 2005. Latitudinal clines in bill length and sex ratio in a migratory shorebird: a case of resource partitioning? Acta Oecologica-International Journal of Ecology 28:33-38.

Nelson, D. A. 1988. Feature Weighting in Species Song Recognition by the Field Sparrow (Spizella pusilla). Behaviour 106:158-182.

Nelson, D. A., and P. Marler. 1990. The perception of bird song and an ecological concept of signal space. Pages 443-478 in Comparative perception: vol. 2, Complex signals (M. A. Berkley, and W. C. Stebbins, Eds.). Wiley, New York.

112

O'Hara, P. D., G. Fernandez, F. Becerril, H. de la Cueva, and D. B. Lank. 2005. Life history varies with migratory distance in Western Sandpipers Calidris mauri. Journal of Avian Biology 36:191-202.

O'Hara, P. D., G. Fernandez, B. Haase, H. de la Cueva, and D. B. Lank. 2006. Differential migration in Western Sandpipers with respect to body size and wing length. Condor 108:225-232.

Panhuis, T. M., R. Butlin, M. Zuk, and T. Tregenza. 2001. Sexual selection and speciation. Trends in Ecology & Evolution 16:364-371.

Patel, R., R. A. Mulder, and G. C. Cardoso. 2010. What Makes Vocalisation Frequency an Unreliable Signal of Body Size in Birds? A Study on Black Swans. Ethology 116:554-563.

Payne, R. B., and D. F. Westneat. 1988. A Genetic and Behavioral Analysis of Mate Choice and Song Neighborhoods in Indigo Buntings. Evolution 42:935-947.

Price, T. 1998. Sexual selection and natural selection in bird speciation. Philosophical Transactions of the Royal Society of London Series B-Biological Sciences 353:251-260.

Price, T. 2007. Speciation in Birds. Roberts and Company Publishers, Greenwood Village, Colorado.

Read, A. F., and D. M. Weary. 1992. The evolution of bird song: comparative analyses. Philosophical Transactions of the Royal Society of London Series B-Biological Sciences 338:165-187.

113

Ryan, M. J., and N. M. Kime. 2003. Selection on long distance acoustic signals. Pages 225-274 in Acoustic Communication (A. M. Simmons, R. R. Fay, and A. N. Popper, Eds.). Springer Verlag, Berlin.

Sandercock, B. K. 1998. Assortative Mating and Sexual Size Dimorphism in Western and Semipalmated Sandpipers. Auk 115:786-791.

Seddon, N. 2005. Ecological adaptation and species recognition drives vocal evolution in neotropical suboscine birds. Evolution 59:200-215.

Seddon, N., C. A. Botero, J. A. Tobias, P. O. Dunn, H. E. A. MacGregor, D. R. Rubenstein, J. A. C. Uy, J. T. Weir, L. A. Whittingham, and R. J. Safran. 2013. Sexual selection accelerates signal evolution during speciation in birds. Proceedings of the Royal Society B: Biological Sciences 280.

Slabbekoorn, H., and T. B. Smith. 2002. Bird song, ecology and speciation. Philosophical Transactions of the Royal Society of London Series B-Biological Sciences 357:493-503.

Soha, J. A., D. A. Nelson, and P. G. Parker. 2004. Genetic analysis of song dialect populations in Puget Sound White-crowned Sparrows. Behavioral Ecology 15:636-646.

Sorjonen, J. 1986a. Factors Affecting the Structure Of Song and the Singing Behavior of some Northern European Passerine Birds. Behaviour 98:286-304.

Sorjonen, J. 1986b. Song Structure and Singing Strategies in the Genus Luscinia in Different Habitats and Geographical Areas. Pages 274-285.

Stein, R. W., G. Fernandez, H. de la Cueva, and R. W. Elner. 2008. Disproportionate bill length dimorphism and niche differentiation in wintering western

114

sandpipers (Calidris mauri). Canadian Journal of Zoology-Revue Canadienne De Zoologie 86:601-609.

Stevens, J. 2002. Applied Multivariate Statistics for the Social Sciences. L. Erlbaum, Mahwah, N.J.

von Frisch, O. 1956. Zur Brutbiologie und Jugendentwicklung des Brachvogels (Numenius arquata L.). Z. Tierpsychol. 13:50-81.

von Frisch, O. 1959. Zur Jugendentwicklung, Brutbiologie und vergleichenden Ethologie der Limicolen. Z. Tierpsychol. 16:545-583.

West-Eberhard, M. J. 1983. Sexual selection, Social Competition, and Speciation. Quarterly Review of Biology 58:155-183.

Wiley, R. H., and D. G. Richards. 1978. Physical Constraints on Acoustic Communication in the Atmosphere - Implications for the Evolution of Animal Vocalizations. Behavioral Ecology and Sociobiology 3:69-94.

Wilson, W. H. 1994. Western Sandpiper (Calidris mauri), The Birds of North America Online. (A. Poole, Ed.). Cornell Lab of Ornithology, Ithaca. http://bna.birds.cornell.edu/bna.

Young, F. W. 2003. ViSta “The Visual Statistics System”. [computer software] [on-line] Available: URL: http://forrest.psych.unc.edu/research/index.html.

115

3.7. Tables

Table 3-1. Means for selected frequency and temporal variables of male songs and alarm calls: Means and standard deviations for each site were calculated from individual male means and are presented for descriptive purposes. We restricted our analysis to the fundamental for the frequency domain.

YK-Delta Nome Wales Kotzebue SONGS N=12 N=14 N=11 N=8 Mean +/- SD Mean +/- SD Mean +/- SD Mean +/- SD Song Duration (ms) 2275 +/- 273 2134 +/- 520 2206 +/- 559 1622 +/- 296 Mean center frequency Song 2907 +/- 144 3166 +/- 119 3063 +/- 111 3224 +/- 127 (Hz) Maximum bandwidth Song 1816 +/- 181 1648 +/- 170 1741 +/- 198 1699 +/- 158 (Hz) Intro Number of notes 9.6 +/- 1.8 9.0 +/- 3.4 7.5 +/- 2.3 5.4 +/- 1.6 Intro Duration (ms) 1377 +/- 183 1487 +/- 423 1425 +/- 363 955 +/- 187 Trill Duration (ms) 893 +/- 123 666 +/- 148 776 +/- 270 660 +/- 159 Trill Initial trill rate (s-1) 120 +/- 10 120 +/- 6 118 +/- 7 116 +/- 9 Trill Final trill rate (s-1) 149 +/- 10 127 +/- 7 135 +/- 12 120 +/- 15

ALARM CALLS N=6 N=7 N=5 N=7 Mean +/- SD Mean +/- SD Mean +/- SD Mean +/- SD Duration (ms) 778 +/- 296 525 +/- 163 429 +/- 24 536 +/- 99 Center frequency (Hz) 3492 +/- 452 3416 +/- 245 3435 +/- 65 3475 +/- 110 90% Bandwidth (Hz) 1146 +/- 220 1031 +/- 247 962 +/- 150 1186 +/- 224 Number of notes 14 +/- 3 10 +/- 3 9 +/- 2 11 +/- 3

116

Peak Center Frequency Frequency Bandwidth Duration Mean +/- SD Mean +/- SD Mean +/- SD Mean +/- SD Songs 7.6 +/- 4.3 7.3 +/- 3.8 16.1 +/- 3.2 24.7 +/- 14.1 Alarm calls 8.0 +/- 0.7 8.1 +/- 1.1 21.8 +/- 1.4 33.4 +/- 4.9

117

3.8. Figures 118

Figure 3-2. Male Western Sandpiper alarm call spectrogram. The spectrogram was generated in Raven 1.4 (Hann window, window size of 312 points with 90% overlap, grid spacing 93.8Hz, 3dB filter bandwidth 221Hz). We removed low- frequency noise and cut spectrograms off at 12.75kHz solely for visualization purposes. We only analyzed the

119 fundamental in the frequency domain, labels indicate alarm call components selected for analysis. For a detailed list and description of measured variables see Table 3-3 in the appendix.

120

Figure 3-3. Spectrograms depicting three male songs (A) and alarm calls (B) from each of the four breeding sites. Spectrograms were generated in Raven 1.4 (Hann window, window size of 312 points with 90% overlap, grid spacing 93.8Hz). We removed noise below 1kHz and cut spectrograms off at 12.75kHz solely for visualization purposes. The examples were chosen to reflect some intra-site variation and are not necessarily representative in every aspect. The first two alarm calls at the Wales site were recorded from the same individual, all other alarm calls and all songs represent different individuals.

Figure 3-4. Individual male principal component scores for song: Scatterplot depicting the first two principal component scores for each male. Our spectrographic analysis yielded 46 frequency and temporal variables which were entered into the PCA. Between 3 and 15 songs were analyzed per male, but only the mean values for each male were entered into the analysis. Symbols represent sites at which males were recorded.

121

122

Figure 3-6. Individual male principal component scores for alarm calls: Scatterplot depicting the first two principal component scores for each male. Our spectrographic analysis yielded 17 frequency and temporal variables which were entered into the PCA. Between 5 and 9 alarm calls were analyzed per male, but only the mean values for each male were entered into the analysis. Symbols represent sites at which males were recorded. Alarm calls from all sites overlap and no separation among sites is evident.

123

Figure 3-7. Mean PC1 scores for male alarm calls at the different breeding sites: Error bars represent 95% Confidence Intervals. Alarm calls did not differ among sites.

124

3.9. Appendix

Usage of a different microphone at the southernmost site (YK-Delta):

Males at the southernmost site in the YK-Delta were recorded with a different microphone than those at the other three sites, which is a potential concern as it is possible for microphones to vary in their frequency response. We attempted to minimize potential microphone artifacts by ensuring that both microphones were of the same make and model. Furthermore, microphones of the chosen model (Sennheiser ME62) show a flat frequency response up to 4kHz, based on manufacturer information (as is typical for omnidirectional microphones). It is therefore very unlikely that one of the microphones would have created an artifact by showing a response peak at a particular frequency (or within a particular frequency range) within the flat response range. In the event of an equipment-based, consistent difference in frequency response, we should also have seen corresponding differences in alarm call frequencies which cover a substantial portion of the frequency range for which there are apparent differences for songs (Appendix Table 3-3). This was not the case, however: YK-Delta song frequency measures were consistently the lowest among sites, while alarm call frequencies clearly did not show this pattern (Appendix Table S3-3). We therefore conclude that equipment artifacts cannot explain the frequency patterns that contributed to the geographical variation in the song- based PC.

125

Table 3-3. Mean values of all song and alarm call variables that were entered into the PCA. Means and standard deviations are given for each breeding site. Within each selection the variable Peak Frequency refers to the frequency with the highest power, Center Frequency to the frequency at which the selection can be divided into two sections with equal amounts of energy, and 90% Bandwidth describes the frequency interval that encompasses 90% of the energy, with 5% of the energy above and below this interval. The song selections i1, i2, i3 and i4 refer to the first, second, third and last introductory note respectively, t1 encompasses the first trill pulse, t2 pulses 1-5, t3 pulses 6-10, t4, t5 and t6 the third-last, second-last and last pulse group of the trill and t7 the last 12 contiguous pulses (see Fig. 3-1). The alarm call selection 1 encompasses the entire call, whereas selections 2, 3 and 4 contain the first, third and last note respectively (see Fig. 3-2).

YK-Delta Nome Wales Kotzebue SONGS N=12 N=14 N=11 N=8 Selection and variable Mean +/- SD Mean +/- SD Mean +/- SD Mean +/- SD i1 Peak Frequency (Hz) 2755 +/- 242 2905 +/- 128 2876 +/- 132 3054 +/- 147 i2 Peak Frequency (Hz) 3316 +/- 169 3459 +/- 119 3450 +/- 82 3495 +/- 169 i3 Peak Frequency (Hz) 3318 +/- 173 3494 +/- 110 3447 +/- 79 3602 +/- 104 i4 Peak Frequency (Hz) 3347 +/- 179 3512 +/- 92 3458 +/- 79 3662 +/- 88 t1 Peak Frequency (Hz) 3395 +/- 145 3590 +/- 86 3553 +/- 82 3670 +/- 132 t2 Peak Frequency (Hz) 3444 +/- 159 3618 +/- 93 3566 +/- 70 3722 +/- 96 t3 Peak Frequency (Hz) 3433 +/- 153 3601 +/- 96 3549 +/- 83 3701 +/- 86 t4 Peak Frequency (Hz) 2392 +/- 205 3041 +/- 250 2780 +/- 322 2977 +/- 336 t5 Peak Frequency (Hz) 2385 +/- 192 2988 +/- 247 2780 +/- 291 3006 +/- 326 t6 Peak Frequency (Hz) 2442 +/- 211 2894 +/- 193 2617 +/- 219 2884 +/- 257 t7 Peak Frequency (Hz) 2418 +/- 197 3062 +/- 334 2844 +/- 406 3064 +/- 290 i1 Center Frequency (Hz) 2581 +/- 242 2743 +/- 100 2718 +/- 153 2849 +/- 174 i2 Center Frequency (Hz) 3151 +/- 195 3225 +/- 130 3203 +/- 90 3266 +/- 180 i3 Center Frequency (Hz) 3203 +/- 193 3297 +/- 115 3259 +/- 90 3394 +/- 122 i4 Center Frequency (Hz) 3263 +/- 189 3367 +/- 107 3326 +/- 92 3502 +/- 85 t1 Center Frequency (Hz) 3364 +/- 148 3534 +/- 83 3506 +/- 80 3612 +/- 119 t2 Center Frequency (Hz) 3375 +/- 162 3540 +/- 83 3501 +/- 68 3628 +/- 84 t3 Center Frequency (Hz) 3374 +/- 169 3528 +/- 97 3496 +/- 70 3617 +/- 85 t4 Center Frequency (Hz) 2426 +/- 185 3002 +/- 241 2780 +/- 296 2963 +/- 308 t5 Center Frequency (Hz) 2413 +/- 185 2970 +/- 238 2749 +/- 280 2976 +/- 307 t6 Center Frequency (Hz) 2389 +/- 181 2863 +/- 192 2584 +/- 228 2843 +/- 228 t7 Center Frequency (Hz) 2432 +/- 188 3030 +/- 279 2810 +/- 352 2987 +/- 278 i1 90% Bandwidth (Hz) 794 +/- 118 782 +/- 109 669 +/- 48 745 +/- 103

126

YK-Delta Nome Wales Kotzebue i2 90% Bandwidth (Hz) 663 +/- 153 653 +/- 92 568 +/- 53 618 +/- 91 i3 90% Bandwidth (Hz) 695 +/- 138 671 +/- 97 579 +/- 67 599 +/- 102 i4 90% Bandwidth (Hz) 675 +/- 153 681 +/- 111 612 +/- 66 637 +/- 116 t1 90% Bandwidth (Hz) 878 +/- 208 814 +/- 90 838 +/- 123 791 +/- 179 t2 90% Bandwidth (Hz) 924 +/- 244 837 +/- 96 849 +/- 129 828 +/- 202 t3 90% Bandwidth (Hz) 924 +/- 256 856 +/- 117 858 +/- 154 787 +/- 239 t4 90% Bandwidth (Hz) 830 +/- 104 836 +/- 115 807 +/- 102 782 +/- 104 t5 90% Bandwidth (Hz) 850 +/- 109 789 +/- 99 838 +/- 99 791 +/- 123 t6 90% Bandwidth (Hz) 785 +/- 106 824 +/- 184 796 +/- 111 797 +/- 164 t7 90% Bandwidth (Hz) 853 +/- 119 876 +/- 103 876 +/- 119 893 +/- 69 i1-t1 90% Bandwidth (Hz) 1494 +/- 211 1450 +/- 139 1384 +/- 154 1424 +/- 176 t1-t6 90% Bandwidth (Hz) 1659 +/- 231 1375 +/- 168 1605 +/- 236 1490 +/- 119 i1 Duration (ms) 288 +/- 65 226 +/- 40 295 +/- 46 258 +/- 51 i2 Duration (ms) 75 +/- 13 131 +/- 72 132 +/- 52 172 +/- 58 i3 Duration (ms) 55 +/- 8 86 +/- 38 88 +/- 33 105 +/- 36 i4 Duration (ms) 43 +/- 7 53 +/- 15 56 +/- 18 64 +/- 17 t2 Duration (ms) 44 +/- 4 45 +/- 3 45 +/- 3 46 +/- 4 t3 Duration (ms) 40 +/- 3 38 +/- 2 40 +/- 2 41 +/- 3 t7 Duration (ms) 81 +/- 6 96 +/- 5 90 +/- 9 102 +/- 12 Song Duration (ms) 2275 +/- 273 2134 +/- 520 2206 +/- 559 1622 +/- 296 Intro Duration (ms) 1377 +/- 183 1487 +/- 423 1425 +/- 363 955 +/- 187 Trill Duration (ms) 893 +/- 123 666 +/- 148 776 +/- 270 660 +/- 159 Intro Number of notes 9.6 +/- 1.8 9.0 +/- 3.4 7.5 +/- 2.3 5.4 +/- 1.6

ALARM CALLS N=6 N=7 N=5 N=7 Selection and variable Mean +/- SD Mean +/- SD Mean +/- SD Mean +/- SD 1 Peak Frequency (Hz) 3678 +/- 442 3608 +/- 257 3596 +/- 104 3685 +/- 223 2 Peak Frequency (Hz) 3398 +/- 519 3482 +/- 261 3428 +/- 128 3547 +/- 233 3 Peak Frequency (Hz) 3622 +/- 487 3568 +/- 274 3536 +/- 75 3757 +/- 232 4 Peak Frequency (Hz) 3478 +/- 440 3445 +/- 263 3443 +/- 113 3437 +/- 123 1 Center Frequency (Hz) 3492 +/- 452 3416 +/- 245 3435 +/- 65 3475 +/- 110 2 Center Frequency (Hz) 3208 +/- 580 3319 +/- 248 3315 +/- 119 3320 +/- 218 3 Center Frequency (Hz) 3492 +/- 475 3431 +/- 267 3426 +/- 65 3546 +/- 153 4 Center Frequency (Hz) 3344 +/- 445 3349 +/- 285 3335 +/- 109 3347 +/- 120 1 90% Bandwidth (Hz) 1146 +/- 220 1031 +/- 247 962 +/- 150 1186 +/- 224 2 90% Bandwidth (Hz) 1103 +/- 162 1080 +/- 258 871 +/- 146 1092 +/- 237 3 90% Bandwidth (Hz) 1023 +/- 200 973 +/- 210 863 +/- 189 1157 +/- 282

127

YK-Delta Nome Wales Kotzebue 4 90% Bandwidth (Hz) 1070 +/- 248 919 +/- 266 839 +/- 120 974 +/- 160 1 Duration (ms) 778 +/- 296 525 +/- 163 429 +/- 24 536 +/- 99 2 Duration (ms) 52 +/- 15 56 +/- 24 59 +/- 28 60 +/- 21 3 Duration (ms) 44 +/- 11 47 +/- 18 43 +/- 11 42 +/- 6 4 Duration (ms) 50 +/- 17 50 +/- 23 40 +/- 10 41 +/- 6 1 Number of notes 14.2 +/- 3.4 10.0 +/- 3.1 9.1 +/- 2.5 11.0 +/- 3.0

128

Table 3-4. Component loadings for the first three principal components for songs and alarm calls. Variables that loaded highest onto PC1 are listed first. Loadings above 0.550 for each component are highlighted in bold, as are the corresponding variables for PC1.

SONGS Component loadings Selection and variable PC1 PC2 PC3 t2 Center Frequency 0.901 -0.144 0.192 t3 Peak Frequency 0.895 -0.102 0.174 t1 Center Frequency 0.891 -0.014 0.184 t2 Peak Frequency 0.888 -0.069 0.192 t3 Center Frequency 0.887 -0.192 0.151 t1 Peak Frequency 0.881 0.043 0.157 t5 Center Frequency 0.861 0.397 -0.114 i4 Peak Frequency 0.86 -0.247 0.145 t4 Center Frequency 0.86 0.396 -0.077 t5 Peak Frequency 0.858 0.405 -0.122 t7 Center Frequency 0.857 0.359 -0.046 i3 Peak Frequency 0.849 -0.316 0.157 t4 Peak Frequency 0.847 0.409 -0.079 t7 Peak Frequency 0.832 0.374 -0.092 i4 Center Frequency 0.786 -0.363 0.326 t6 Center Frequency 0.765 0.531 -0.132 i1 Center Frequency 0.763 -0.193 0.1 i1 Peak Frequency 0.752 -0.22 0.082 t6 Peak Frequency 0.751 0.547 -0.099 Trill Duration -0.726 -0.244 0.166 t7 Duration 0.715 0.322 -0.021 i3 Center Frequency 0.709 -0.414 0.395 i2 Peak Frequency 0.687 -0.397 0.265 Song Duration -0.617 -0.218 0.303 i2 Center Frequency 0.551 -0.421 0.509 Intro Number of notes -0.454 0.013 0.606 t3 90% Bandwidth -0.45 0.597 0.224 i2 Duration 0.448 -0.012 -0.631 i3 Duration 0.409 -0.059 -0.637 Intro Duration -0.408 -0.123 0.313 t2 90% Bandwidth -0.396 0.614 0.176 t1 90% Bandwidth -0.394 0.612 0.171

129

SONGS Component loadings Selection and variable PC1 PC2 PC3 t1-t6 90% Bandwidth -0.357 -0.394 0.49 t7 90% Bandwidth 0.351 0.144 0.55 t2 Duration 0.348 0.123 0.022 i3 90% Bandwidth -0.346 0.636 0.29 i2 90% Bandwidth -0.332 0.677 0.145 i4 Duration 0.319 -0.155 -0.616 i4 90% Bandwidth -0.27 0.772 0.107 i1 Duration -0.225 -0.331 -0.271 i1-t1 90% Bandwidth -0.212 0.157 0.288 t4 90% Bandwidth 0.17 0.194 0.641 t6 90% Bandwidth 0.148 0.419 0.41 t3 Duration 0.129 0.105 0.139 i1 90% Bandwidth -0.128 0.001 0.43 t5 90% Bandwidth 0.033 0.163 0.596

ALARM CALLS PC1 PC2 PC3 1 Center Frequency (Hz) 0.955 0.057 -0.162 1 Peak Frequency (Hz) 0.942 0.042 0.035 3 Center Frequency (Hz) 0.923 0.177 -0.157 4 Center Frequency (Hz) 0.904 0.108 -0.125 2 Center Frequency (Hz) 0.873 0.262 -0.154 2 Peak Frequency (Hz) 0.866 0.217 -0.101 3 Peak Frequency (Hz) 0.85 0.28 -0.007 4 Peak Frequency (Hz) 0.796 0.209 -0.027 1 90% Bandwidth (Hz) 0.544 -0.479 0.614 2 90% Bandwidth (Hz) 0.514 -0.36 0.546 4 90% Bandwidth (Hz) 0.42 -0.374 0.348 3 90% Bandwidth (Hz) 0.365 -0.482 0.647 1 Number of notes 0.047 -0.368 -0.151 2 Duration (ms) -0.165 0.836 0.155 3 Duration (ms) -0.165 0.732 0.617 4 Duration (ms) -0.21 0.567 0.726 1 Duration (ms) -0.264 0.012 0.311

130

Chapter 4. Inter- and intra-individual variation in the vocalizations of male Western Sandpipers

4.1. Abstract

Acoustic signals can encode multiple messages, and may inform receivers about such factors as a signaller’s location, species, group or individual identity, sex, size, fitness or condition, age, social status and motivational or hormonal states. As a consequence, even a single type of vocalization may show significant degrees of intraspecific variation, e.g. among populations, individuals or even in the form of temporal variation for the same individual. Inter-individual variation may facilitate discrimination among familiar and unfamiliar individuals, or individual identification which may confer fitness benefits to both sender and receiver, e.g. in the context of parent-offspring recognition, or discrimination between established neighbours and unfamiliar intruders. In certain cases it may also provide a basis for sexual selection. One potential source of inter-individual variation in vocalization frequencies are differences in body size. Temporal variations in the vocalizations of an individual, on the other hand, may reflect changes in e.g. behavioural context, condition, status or developmental stage. We investigated several aspects of intraspecific variation in male songs and alarm calls in the Western Sandpiper (Calidris mauri), a shorebird, and thus a species whose vocalizations are presumed to be primarily innate. These aspects included vocal individuality, potential effects of body size on vocalization frequencies and reproductive status. Discriminant function and frequency correlation analyses suggested the presence of vocal individuality. Furthermore, our data suggested that body size may contribute to

131

inter-individual variation as song frequency characteristics showed a significant negative correlation with body size (r=-0.372, P=0.044) and alarm call frequencies a negative trend (r=-0.32; P=0.059). In contrast to body size, reproductive status appeared to primarily influence the number and length of certain song components as well as bandwidth characteristics but not song frequencies. Our study sheds some light onto sources of variation in Western Sandpiper songs and alarm calls. Nevertheless, our knowledge about what influences shorebird vocalizations is far from complete and further research into the effects and communicative value of such factors as morphology, condition and behavioural contexts is needed.

4.2. Introduction

A wide variety of species across animal taxa use sound to communicate over medium to large distances. Their acoustic signals can carry a wide range of possible information about the signaller such as location, species, group or individual identity, sex, size, fitness or condition, age, social status and motivational or hormonal states (Hill and Lein 1987; Lampe and Espmark 1994; Galeotti et al. 1997; Ritchie et al. 1999; Saarikettu et al. 2005; Theis et al. 2007; Tyack 2008; Araya-Ajoy et al. 2009; Geberzahn et al. 2009; Koren and Geffen 2009; Rek and Osiejuk 2010; Charlton et al. 2012b; Bolt 2013; Keen et al. 2013; Spehar and Di Fiore 2013; Vehrencamp et al. 2013), frequently providing multiple cues simultaneously or sequentially (Falls 1982; Masters et al. 1995; Galeotti et al. 1997; Nelson and Poesel 2007; Koren and Geffen 2009; Vehrencamp et al. 2013). The multidimensional nature, long range and relative flexibility of acoustic communication may make it particularly well-suited at conveying these multiple messages which may be encoded in various ways in the frequency and/or temporal domain, or may be communicated through amplitude-modulation differences (Gerhardt 1992; Searby et al. 2004; Castellano and Rosso 2007; Catchpole and Slater 2008; Charrier et al. 2009; Koren and Geffen 2009).

132

As a consequence, a single vocalization type (e.g. the song of a bird) may be highly variable even within a species. Such intraspecific variation may reflect geographical differences, but may also arise from within-population differences between the sexes and/or individuals and even through temporal variation on the level of an individual (Price 2007; Theis et al. 2007; Catchpole and Slater 2008; Vannoni and McElligott 2009; Castellano and Gamba 2011).

Inter-individual variation may facilitate individual identification which – depending on signal type and context – may confer fitness benefits to both sender and receiver (Falls 1982; Tibbetts and Dale 2007; Catchpole and Slater 2008). Vocal individuality and/or individual recognition has been demonstrated for cases with obvious fitness benefits, such as mate or parent-young recognition in colonially-breeding species or those with precocial young (Baker 1982; Falls 1982; Stoddard and Beecher 1983; Beecher et al. 1985; Robisson et al. 1993; Charrier et al. 2003; Clark et al. 2006; Blumenrath et al. 2007; Knörnschild et al. 2013). Discrimination between known neighbours and unfamiliar intruders has also been well-established for many territorial species (Falls 1982; Ydenberg et al. 1988; Lovell and Lein 2004a; Hardouin et al. 2006; Skierczyński et al. 2007; Tripovich et al. 2008; Feng et al. 2009; Mager et al. 2010; Wilson and Mennill 2010; Wei et al. 2011; Budka and Osiejuk 2013; Draganoiu et al. 2014). While the degree of individual distinctiveness and its associated fitness benefits may vary across species and vocalization types, individually distinctive vocalizations appear to be a wide-spread, possibly universal feature in vocalizing species (Falls 1982; Lambrechts 1996; Terry et al. 2005; Catchpole and Slater 2008). Genetic differences as well as developmental and environmental influences likely contribute towards vocal individuality (Falls 1982; McGregor 1993; Terry et al. 2005; Forstmeier et al. 2009).

In addition to aiding in individual recognition, inter-individual variation in the features of certain vocalization types (e.g. songs) may also provide a basis for sexual selection and/or facilitate assortative mating (Darwin 1871; Searcy and Andersson 1986;

133

Catchpole 1987). The features used for signalling identity vs. signalling e.g. quality in a mating context are not necessarily identical – in fact, theory predicts they should not be the same based on the differing modes of selection (diversifying vs. directional) that theoretically underlie these two functions (Dale et al. 2001). Sexual selection, for instance, may be based on certain call or song characteristics that constitute honest signals of desirable qualities such as learning or performance abilities, or physical characteristics such as body size (Giacoma et al. 1997; Nowicki et al. 1998; Nowicki et al. 2002; Ballentine et al. 2004; Nolan and Hill 2004; Nowicki and Searcy 2004; Podos et al. 2004; Boogert et al. 2008; Bradbury and Vehrencamp 2011; Charlton et al. 2012a).

A negative relationship between vocalization frequencies and body size has long been postulated (Konishi 1970; Morton 1977; Lambrechts 1996; Fitch and Hauser 2003; Bradbury and Vehrencamp 2011). The size of a sound producing organ determines which wavelengths can be produced efficiently and can be expected to correlate with measures of overall body size, thereby inversely linking body size with vocalization pitch (Konishi 1970; Bennet-Clark 1998; Fletcher 2004; Bradbury and Vehrencamp 2011). Empirical support for this frequency allometry hypothesis stems from many major taxa, including birds. While this inverse relationship – with a few exceptions – appears to be fairly consistent when species or higher-level taxa are compared (e.g. Wallschläger 1980; Duellman and Pyles 1983; Ryan and Brenowitz 1985; Wiley 1991; Vaughan et al. 1997; Jurisevic and Sanderson 1998; Tubaro and Mahler 1998; Matthews et al. 1999; Bertelli and Tubaro 2002; Fletcher 2004; Seddon 2005; Mahler and Gil 2009; Martin et al. 2011; Stoffberg et al. 2011; but see Cardoso and Mota 2007), studies investigating within- species patterns in birds have produced surprisingly mixed results (Schubert 1976; Shy 1983; Barbraud et al. 2000; Laiolo et al. 2004; Irwin et al. 2008; Patel et al. 2010; Hall et al. 2013). Individual flexibility and intra-individual variation in vocalization frequencies may, for instance, surpass any effects imposed by the relatively small intraspecific body size differences and hence may obscure the latter (Cardoso et al. 2008; Patel et al. 2010).

134

As with inter-individual variation there are many possible sources of intra- individual variation. Temporal variation in the vocalizations of an individual may reflect changes in e.g. behavioural context, condition, status or developmental stage (Miller 1984; Hill and Lein 1987). Nuptial vocalizations, for instance, may change even over the course of a single breeding season (Catchpole 1983; Rios-Chelen et al. 2005; Catchpole and Slater 2008; Vannoni and McElligott 2009).

While a variety of aspects of inter- and intra-individual variation have been studied extensively in songbirds, the respective data in other bird taxa, and particularly in shorebirds, is far from comprehensive. Song learning and plasticity in songbirds appears to be responsible for a significant portion of such variation and may mask other sources (Osiejuk et al. 2005; Cardoso et al. 2008; Patel et al. 2010). Taxa with primarily innate vocalizations and differing vocal production mechanisms, such as shorebirds, can thus be expected to show patterns of variation that may differ from those of oscines and other song-learning taxa (Lovell and Lein 2004b; Odom and Mennill 2012).

Our study aims to investigate inter- and intra-individual sources of variation for two types of vocalizations – songs and alarm calls – of male Western Sandpipers (Calidris mauri). Male Western Sandpiper songs appear to function in mate attraction and territory or resource defense while alarm calls may alert chicks and/or mating partners to the presence of a predator and focus the attention of a predator towards the parent – away from the nest or young (Brown 1962; Holmes 1973; Lanctot et al. 2000; Johnson et al. 2008; Chapter 5). Both vocalization types are likely innate, yet they differ in their geographical patterns of variation.

Male songs vary across the breeding range, particularly with regard to their frequency characteristics (Chapter 3). A variance components analysis further indicated that a substantial portion (24%) of the overall variance is attributable to inter-individual differences within sites and that this component exceeds within-male variation (Chapter

135

3). The underlying ultimate and proximate causes of this geographical and inter- individual variation remain undetermined, but morphological characteristics might contribute to song frequency differences between sites as well as between individual males (Chapter 3). A moderate amount of the frequency variation in contact calls across shorebird species is explained by body size (24%) or mass (20%) and differences in call frequencies among populations may be related to body size differences in some species (Miller 1986; Bried and Jouventin 1997; Douglas and Conner 1999). However, no published studies confirm such a relationship on an individual level within a species.

Western Sandpiper alarm calls, on the other hand, do not appear to differ among breeding sites. Furthermore, the degree to which they may differ among individuals within a site remains unclear (Chapter 4).

In the current chapter we further explore variation and its possible sources in both types of vocalizations. We investigate whether the inter-individual variation among males is sufficiently large to make males individually identifiable based on the vocal characteristics of their songs or alarm calls. Concordantly, we test whether the vocal characteristics of these males show consistencies in the frequency domain across vocalization types: specifically, do males that sing at lower fundamental frequencies have lower-pitched alarm calls than males that sing at higher frequencies? We also investigate if and to what degree individual frequency differences can be attributed to differences in body size among individuals within and among sites. Finally, we explore the effects of breeding stage as a potential source of intra-individual variation for male songs.

Our study will help to further illuminate the types of information that may be encoded in shorebird vocalizations, shed light on what factors may drive geographical variation and act as an important stepping stone for further studies which may explore if and how Western Sandpipers and other shorebirds utilize the potential sources of information contained in the vocalizations of their conspecifics.

136

4.3. Methods

We obtained recordings of male songs and alarm calls at four breeding sites in western Alaska: in the Yukon-Kuskokwim-Delta (YK-Delta; 61°21'N, 165°07'W; June 11-15, 2009) with a Marantz PMD 660 solid state recorder and a Sennheiser ME62 microphone mounted inside a 56cm parabolic dish and near Nome (64°26'N, 164°56'W; May 21-26, June 07, 2009; and May 21 – July 02, 2010), Wales (65°36'N, 168°05'W; May 29 – June 04, 2009) and Kotzebue (66°50'N, 162°34'W; June 11-18, 2009) with a Marantz PMD 670 solid state recorder and a Sennheiser ME62 inside a 56cm parabolic dish (see Chapter 3). We recorded in wave format with a sampling rate of 48kHz and bit depth of 16 at distances of 5 to 35m (estimated to the nearest 5m). In one instance a male may have ranged to a distance of up to 105m.

We analyzed the fundamental frequency of both vocalization types in Raven (1.4; Bioacoustics Research Program 2011; Hann window of 312 points with 90% overlap, grid spacing 46.9Hz). We measured the number of notes, duration, 90% bandwidth and peak and center frequencies of selected song and alarm call components (for further details on the analysis see Chapter 3, especially Fig. 3.1 and 3.2 and Appendix Table 3-3 for a detailed description and visualization of the selected components and measurements for each vocalization type). For the frequency-specific analyses of songs we included peak and center frequency variables representing different song components (see Appendix Table 4-3 for a list of variables). In the case of alarm calls we chose the overall center frequency (measured across the entire call) as a representative frequency measure as it was highly correlated with all other frequency measures (Pearson’s correlation coefficient r=0.78-0.95, P<0.001) and frequency measures varied less across call components.

Principal Component Analyses were conducted in SYSTAT (13.1; SYSTAT Software, Inc.), associated Parallel Analyses in ViSta (Young 2003) and all other

137

statistical analyses in SPSS Statistics (v17.0, SPSS Inc.). Pearson’s correlation coefficient was used in all correlation analyses.

4.3.1. Vocal individuality – evidence for identity signalling:

To investigate whether individuals can be identified based on the vocal characteristics of each of the two vocalization types, we conducted canonical discriminant function analyses (DFA). We restricted each analysis to data from a single breeding site and year, to avoid artificially inflating inter-individual differences through inter-site (see chapter 3) or potential inter-year variation. For both questions we selected the site with the largest sample size: Nome for songs and Kotzebue for alarm calls. We excluded one male with an insufficient sample size from the song analysis leaving us with 125 songs from 13 males and 55 alarm calls from 7 males that entered the analysis.

To reduce the number of variables, which was necessary to allow the DFA to progress, and to prevent collinearity issues, as some of the original variables were highly correlated, we first conducted Principal Component Analyses (PCA) using individual song attributes. Missing values were replaced with the mean (<0.4% of the data). Mean imputation is one of the most commonly used methods to replaces missing values and our percentage of missing values is below 1% which is considered trivial (Acuña and Rodriguez 2004). Therefore this should not constitute a problem for our statistical analyses. We included all Principal Components (PCs) with eigenvalues > 1 (10 for songs and 6 for alarm calls; for loadings see Appendix Table 4-1) in the DFAs which allowed us to reduce the number of variables while retaining a large proportion of the total variation (79.4 and 82.4% respectively). For the DFA we used a stepwise model (threshold of F=3.84 for entry and F=2.71 for removal of a variable) with prior probabilities calculated according to individual sample size. As covariance matrices were not equal we chose the separate groups covariance matrix option which meant a leave- one-out cross-validation could not be conducted. According to Naguib et al. (2001),

138

citing Pimentel and Frey (1978), the over-estimation of discriminatory power this may result in should be negligible for males for which multiple vocalizations are included in the analysis.

Given our small sample sizes, sample splitting was also not an option. For songs, the first six PCs and PC8 entered the model under the stepwise procedure, yielding seven discriminant functions (for variable contributions to each function see Appendix Table 4- 2). In the case of alarm calls the first five PCs entered and five functions discriminated among individuals.

We also investigated whether individual fundamental frequency characteristics were correlated among the two vocalization types using data from all four breeding sites. We entered mean song frequency measures (see above) for each individual into a PCA, conducted a Parallel Analysis in ViSta (Young 2003) and in accordance with the results from this analysis retained the first PC. This PC (song frequency PC1) had solely positive loadings and accounted for 64.8% of the frequency variation (for component loadings see Appendix Table 4-3a). We then conducted a correlation analysis using song frequency PC1 and the mean alarm call center frequency of all males from which both types of vocalizations were recorded (n=21).

4.3.2. Correlation of body size and vocal frequency characteristics:

We tested whether frequency characteristics of either type of vocalization correlated with body size across individual males from the different breeding sites (songs: n=22; alarm calls: n=25) and within the site with the highest sample size (Nome; songs: n=11; alarm calls: n=10).

As in the analysis above, we conducted a PCA of song frequency measures and used song frequency PC1, which accounted for 65.9% of the variation, as representative frequency measure for songs (for loadings see Appendix Table 4-3b) and overall center

139

frequency for alarm calls. Where the correlation was significant for PC1, we also performed body size correlation tests for each of the underlying frequency variables (univariate correlations) to provide a more detailed illustration of the strength of the effect on the frequencies of different song components. As this represents a follow-up on a significant multivariate test and has primarily descriptive purposes we did not adjust P- values for multiple comparisons.

To obtain a measure of body size, we caught males in walk-in traps that were placed over active nests. We measured culmen and full tarsus length (to the nearest 0.1mm), wing length (straightened and flattened wing chord, to the nearest 0.5mm) and body mass (to the nearest 0.5g) and conducted a PCA on these variables. All variables loaded positively onto the first PC (morphology PC1) which accounted for 38.0% of the variation (Appendix Table 4-4). Additional PCs were characterized by mixtures of positive and negative loadings and hence appeared to reflect shape rather than overall body size. We therefore retained solely PC1 as a representative measure of body size.

4.3.3. Reproductive status and song characteristics:

We investigated whether song characteristics were influenced by male reproductive status. We differentiated among males recorded prior to incubation (pre- incubating males), during incubation (incubation males) and during chick-rearing (chick- rearing males). We restricted our analysis to males recorded at the Nome breeding site – the site with the largest sample size and the only site where males were recorded at all three reproductive stages. Males that were recorded early in the breeding season, had no apparent association with a nest and exhibited no distraction behaviour were presumed to be in the pre-incubation stage. Males in this group may have been paired or unpaired. Males that were designated as incubating males actively attended a nest containing eggs, and chick-rearing males were leading and/or brooding young. To ensure comparability with the among-site analysis in chapter 3, we retained the component loadings from the

140

song PCA in chapter 4 and calculated PCs accordingly. As in chapter 3, we retained the first three PCs (for loadings see chapter 3) which we compared among reproductive stages using ANOVA, following up on significant results with post-hoc Tukey's HSD tests.

4.4. Results

4.4.1. Vocal individuality – evidence for identity signalling:

The discriminant function analysis correctly assigned 92.8% of the 125 songs to the 13 males recorded at the Nome breeding site (Fig. 4-1a). For alarm calls, no classification errors occurred for the seven males at the Kotzebue site, i.e. 100% of the 55 calls were correctly assigned (Fig. 4-1b). We selected the Kotzebue site for the alarm call DFA based on its larger sample size, but an exploratory DFA of five males in Nome also yielded a high degree of accurate classification with only a single misclassified call. Our correlation analysis detected a strong congruence between alarm call center frequency and song frequency PC1 across the four sites (Fig. 4-2a, n=21, r=0.587, P=0.003).

4.4.2. Correlation of body size and vocal frequency characteristics:

Song frequency PC1 showed a moderately strong, inverse relationship with body size, as indexed by morphology PC1, across individual males from the different breeding sites (Fig. 4-3a, n=22, r=-0.372, P=0.044). When males from only one breeding site (Nome) were considered, the trend was also negative, but not significantly so (n=11, r=- 0.128, P=0.354). Significant univariate correlation coefficients across sites ranged from r=-0.385 to r=-0.472 and variation in body size did not appear to affect frequencies in the first introductory note and the terminal portion of the trill (Appendix Table 4-5).

141

Alarm call center frequency was also negatively correlated with morphology PC1 across breeding sites, but this relationship was weaker and only approached significance (Fig. 4-3c, n=25, r=-0.32; P=0.059). Within a single breeding site (Nome), there was a weak negative trend that was not significant (Fig. 4-3d, n=10, r=-0.265, P=0.229).

The negative correlations between body size and frequency did not appear to be driven by differences in the amount of body fat deposits, as there was no apparent relationship between fat scores and song or alarm call frequencies (Spearman’s Rank Correlation, P=0.433 and P=0.293).

4.4.3. Reproductive status and song characteristics:

We did not detect differences among the three different reproductive stages with regard to the characteristics of male songs represented by PC1 and PC2 (n=30; PC1: F=1.52, df=2 and 27, P=0.236; PC2: F=0.11, df=2 and 27, P=0.901). PC3 values, on the other hand, were higher for pre-incubation than chick-rearing males (Fig 4-4, n=30, F=5.5, df=2 and 27, P=0.01; Tukey post- hoc test significant at P=0.008 for pre- incubation vs. chick-rearing). This principal component primarily represented the variation in the number of introductory notes, their duration (inverse relationship) and the bandwidth characteristics of terminal trill elements (for detailed information on component loadings see chapter 3).

4.5. Discussion

4.5.1. Vocal individuality – evidence for identity signalling:

Our study demonstrates that it is possible to discriminate among Western Sandpiper males within a breeding site based on individual vocal characteristics that are evident in both songs and alarm calls. While it is possible that our assignment reliability

142

is an overestimate, as we could not perform a leave-one-out cross-validation, the clustering of each individual’s vocalizations (see Fig. 4-1) does suggest individuality. Such individuality of acoustic signals appears to be a wide-spread phenomenon, even across species with primarily innate vocalizations (Falls 1982; Lambrechts 1996; Terry et al. 2005; Catchpole and Slater 2008; Foote et al. 2013). It has been demonstrated for mammals, birds, fish, amphibians and insects across a variety of functionally different signal types (Baker 1982; Bretagnolle 1996; Crawford et al. 1997; Barbraud et al. 2000; Naguib et al. 2001; Darden et al. 2003; Searby et al. 2004; Dragonetti 2007; Policht et al. 2009; Melendez and Feng 2010; Koren and Geffen 2011; Digweed et al. 2012; Zhang et al. 2012; Eckenweber and Knörnschild 2013; Foote et al. 2013; Keen et al. 2013; Pettitt et al. 2013). Given its prevalence, vocal individuality likely carries advantages for both sender and receiver in many cases, though its importance and specific function may vary across species, signal type and context (Falls 1982; Marler 2004; Tibbetts and Dale 2007; Catchpole and Slater 2008; Keen et al. 2013).

For male Western Sandpipers several advantages appear plausible. Males establish and defend breeding territories and physical clashes are common (Brown 1962; Holmes 1971, 1973). Consequently, vocal individuality and individual recognition should offer the same fitness benefits that have been postulated for songbirds and other territorial species. If males are individually recognizable by their songs, established neighbours can be distinguished from unknown intruders that likely present a greater threat, avoiding unnecessary and potentially costly fights (Falls 1982; McGregor 1993; Tibbetts and Dale 2007). Males may further benefit from vocal individuality if females visit and assess several males prior to choosing a mate, which they then need to recognize and locate again (Thom and Dytham 2012). Vocal individuality may also facilitate recognition among mated pairs, thereby aiding in the coordination of incubation and parental care duties, territorial defense and predator distraction (Falls 1982; Clark et al. 2006). Finally, vocal individuality in both songs and alarm calls may aid in maintaining family unit cohesion after the chicks have hatched. Western Sandpiper chicks are highly precocial,

143

depart the nest within hours of hatching and are capable of travelling several hundred meters a day (Holmes 1972, 1973; Ruthrauff and McCaffery 2005). Adults nevertheless provide crucial parental care in the form of brooding, guidance towards feeding areas and predator protection, and chicks adjust their behaviour in response to adult vocalizations (Brown 1962; Holmes 1971, 1972, 1973; Johnson 2006; Johnson et al. 2008). The high mobility of the chicks and the movements of family units out of nesting territories and towards and among feeding areas increase the likelihood of encounters with other individuals or family units during which foreign chicks may be attacked (Johnson and McCaffery 2004; Ruthrauff and McCaffery 2005; Johnson et al. 2008). Recognizable features in parental vocalizations could enable chicks to follow the guidance of the correct parent(s) during such encounters, thereby aiding in the cohesion of family units and reducing the likelihood of such attacks.

Given the evidence from other taxa (e.g. Falls 1982; Bretagnolle 1996; Stoddard 1996; Charrier et al. 2003; Searby et al. 2004; Clark et al. 2006; Blumenrath et al. 2007), it appears highly likely that Western Sandpipers and other shorebird species signal identity and/or are able to recognize other individuals in some or all of the abovementioned contexts. Data on inter-individual variation in shorebird vocalizations is scarce, but what little data is available – from parental calls of Dunlin (Calidris alpina) to contact or advertisement calls of Wilson’s Phalaropes (Phalaropus tricolor), American Woodcocks (Scolopax minor), Piping Plovers (Charadrius melodus) and Least Sandpipers (Calidris minutilla) – suggests that vocal individuality may be a common feature for shorebirds just as it is for many other highly vocal taxa (Howe 1972; Beightol and Samuel 1973; Baker 1982; Miller 1986; Sung and Miller 2007). However, not all vocalization types necessarily provide reliable identity cues or are actually used for recognition by other individuals as evidenced by the apparent inability of adult Stone Curlews (Burhinus oedicnemus) to discriminate between the calls of their own and those of foreign young (Dragonetti et al. 2013). Consequently, playback experiments need to be conducted to test whether shorebirds in fact are able to recognize or discriminate among

144

other individuals based on their vocal signatures and under what circumstances they utilize this information. So far, the only evidence for this ability has been found in Western Sandpipers: chicks appear to respond more strongly to parental than to non- parental vocalizations in this species (Johnson et al. 2008). To our knowledge, individual vocal recognition has not been demonstrated in any other shorebird species or for any of the other abovementioned contexts, though individual recognition or neighbour-stranger discrimination by visual means in Ruddy Turnstones (Arenaria interpres) and Ruffs (Philomachus pugnax) suggests that shorebirds possess the required general cognitive abilities (Whitfield 1986, 1988; D. Lank, unpubl. data).

Consequently, which vocal features may be important for individual recognition processes remains obscure. Studies in other taxa have shown that characteristics in any of the vocal dimensions (frequency, time or amplitude) or combinations thereof may provide essential identity cues to conspecifics and that the parameters that are primarily involved may vary across species, vocalization types and even among sexes (e.g. Bretagnolle 1996; Searby et al. 2004; Vignal et al. 2004; Wiley 2005; Berg et al. 2011; Dentressangle et al. 2012; Odom et al. 2013). The multivariate, multi-level nature of our analysis of Western Sandpiper song and alarm calls renders it difficult to determine the relative importance of the various univariate measures. However, both frequency and temporal parameters entered the model and several discriminant functions were required to achieve a high rate of correct assignments. We therefore hypothesize that several parameters may be required to successfully identify other individuals in this species and that both frequency and temporal parameters may be involved. Experimental tests – e.g. using carefully manipulated playbacks – could shed further light on the relative importance of frequency and temporal characteristics in this process.

We expect, however, that characteristics in the frequency domain would be more closely linked to the physical characteristics of the sound production system and hence will show stronger consistencies across different vocalization types. Our study showed

145

that, for Western Sandpiper males, the fundamental frequencies of two functionally and structurally different types of vocalizations were strongly correlated. Both of these vocalization types are likely innate, as shorebirds do not appear to learn their calls or songs (von Frisch 1956; von Frisch 1959; Miller et al. 1983; Miller 1986; Douglas 1996; Miller 1996). It therefore seems likely that inter-individual variation in the vocal apparatus, e.g. its size, and its effect on fundamental frequencies are sufficiently large to create consistent differences in vocal parameters among individuals.

4.5.2. Correlation of body size and vocal frequency characteristics:

Fundamental frequencies were negatively correlated with overall body size in Western Sandpipers. This relationship is further corroborated by the observation that females, the larger sex, sing at lower fundamental frequencies than males (unpubl. data). For males, our body size index accounted for 10 to 14% of the overall variation in fundamental frequency parameters, signifying a moderate effect of body size on vocalization frequencies. However, our body size index can only serve as an approximation of overall body size and may well be an imperfect representation of the size of the vocal apparatus which most directly affects frequency production (Handford and Lougheed 1991; Fitch and Hauser 2003; Bradbury and Vehrencamp 2011; Martin et al. 2011). We chose to include mass in our multivariate measure of body size as it can be considered the best univariate representation of overall body size (Rising and Somers 1989). Mass may, however, be variable over time as more or less fat is deposited. Nonetheless, we are confident that variation in fat deposits did not unduly influence our results as we included several structural measures in our multivariate body size score. Furthermore, all of our birds were caught during incubation and most had similar fat scores in the lower mid-range (2-3 on a scale of 0-7). Only one bird had a higher fat score of 4 and two (alarm calls) or three (songs) birds had a low fat score of 1. Furthermore, there was no significant relationship between fat scores and vocalization frequencies.

146

Therefore, mass differences due to differing fat deposit amounts (rather than structural body size) do not appear to be responsible for the observed correlation.

For alarm calls the negative relationship between frequency and body size only approached significance, and might be slightly weaker than it is for song frequencies overall. Furthermore, our univariate correlation tests suggested that the effect of body size on frequency may be even stronger for some of the song components. For some species, body size may primarily constrain elements that are particularly difficult to perform, such as high-rate trills or particularly low-pitched elements (Konishi 1970; Lambrechts 1996; Bradbury and Vehrencamp 2011; Hall et al. 2013). For Western Sandpipers, songs not only show greater frequency differences among the different song elements than alarm calls, but, as a sexually selected signal, are more likely to contain elements that are challenging to produce and thus may act as performance indicators. A stronger tie of such elements with body size is possible and song frequencies of the different song components did indeed appear to differ in the strength of the observed correlation, with body size explaining up to 22% of the variation in some components while having no detectable effect on others. We could not, however, make out any obvious patterns that would offer a simple explanation. Lower frequencies, for instance, did not appear to be influenced by body size and no obvious differences for trill and introductory note frequencies was apparent. Thus, it appears that body size has a moderate effect on vocalization frequencies in Western Sandpipers, with a somewhat stronger effect on the frequencies of some song elements.

Even with a moderately strong effect, our result is notable as it demonstrates allometry of vocalization frequencies in a shorebird at the level of individual males. We are not aware of any other shorebird species for which the body size-frequency relationship has been studied on an individual level, with the exception of Pectoral Sandpipers (Calidris melanotos) for which male hooting call frequencies are correlated with body size (W. Forstmeier and B. Kempenaers, pers. communication). These hooting

147

calls may represent a somewhat special case, as their frequency is unusually low and their production appears to be closely tied to specialized throat sacs (Miller et al. 1988; Farmer et al. 2013). Comparisons of population means of frequency and morphometric measurements in Least Sandpipers (Calidris minutilla) and Black-faced Sheathbills (Chionis minor) suggest that a negative correlation between body size and frequency may be present in other shorebirds and across different vocalization types (Miller 1986; Bried and Jouventin 1997). In how far this relationship extends to the individual level remains unclear however.

Even among other bird taxa, few studies have been able to demonstrate inverse body size frequency relationships within species. While a few intraspecific studies found negative correlations, others could only detect a weak effect, failed to demonstrate any such relationship or found patterns that are inconsistent with, even opposite to what would be predicted under a frequency allometry hypothesis (e.g. Schubert 1976; Shy 1983; Barbraud et al. 2000; Laiolo et al. 2004; Irwin et al. 2008; Patel et al. 2010; Hall et al. 2013). These incongruities stand in contrast to the relatively robust findings in across species comparisons (Wallschläger 1980; Ryan and Brenowitz 1985; Wiley 1991; Badyaev and Leaf 1997; Jurisevic and Sanderson 1998; Tubaro and Mahler 1998; Bertelli and Tubaro 2002; Seddon 2005; Mahler and Gil 2009; Martin et al. 2011; Hall et al. 2013; but for exceptions see Laiolo and Rolando 2003; Cardoso and Mota 2007). Several possible explanations have been offered for cases of missing or weak intraspecific evidence for inverse frequency-body size relationships. For instance, syrinx size may not be sufficiently correlated with body size or the morphological variation within a species may not be large enough to have a detectable effect on vocalization frequencies, particularly when only a single population and/or one sex is considered (Handford and Lougheed 1991; Fitch and Hauser 2003; Cardoso et al. 2008; Patel et al. 2010). Moreover, any such effect should be even harder to detect if a species is capable of considerable frequency modulation – an easy feat for most birds (Cardoso et al. 2008; Patel et al. 2010). Songbirds in particular, appear to have great control over the

148

frequencies of their vocalizations, more so than other bird taxa (Cardoso et al. 2008). The resulting greater range and variability of oscine vocalization frequencies may help to obscure possible allometric relationships (Cardoso et al. 2008). It also stands to reason that this might particularly affect learned vocalizations (e.g. songs), as template-matching may further mask any body size effects (Osiejuk et al. 2005; Patel et al. 2010). Song learning, therefore, offers a further possible explanation for some of the inconsistent findings in oscine species in particular. Which frequency measures are taken may also influence results, as dominant or low frequencies or those present in components that are difficult to perform may experience greater body size constraints than other frequency components (Konishi 1970; Bowman 1979; Ryan and Brenowitz 1985; Lambrechts 1996; Patel et al. 2010; Hall et al. 2013).

In shorebirds, vocalizations appear to be primarily innate (von Frisch 1956; von Frisch 1959; Miller et al. 1983; Miller 1986; Douglas 1996; Miller 1996), and changes in vocalization characteristics would be expected to be primarily genetically driven (Price 2007); hence we would expect to see an overall more consistent intraspecific link between body size and vocalization frequencies than in oscines. Moreover, a study on the mechanisms of sound production in ring doves suggests that doves experience stronger anatomically-based constraints on the range of their frequency modulation than songbirds (Beckers et al. 2003). These constraints likely also apply to other non-oscines with lateral tympaniform membranes, a feature that is present in shorebirds (Brown and Ward 1990; Beckers et al. 2003). As mentioned above, data on frequency allometry in shorebirds are scarce but all available studies, including ours, suggest an existing relationship to body size and, to our knowledge, there are no studies that failed to find a negative correlation. It is therefore possible that vocalization frequencies constitute a more reliable signal of body size in shorebirds than they do in songbirds. Future studies will have to show whether frequency parameters in shorebirds indeed act as honest signals of body size and are perceived as such by conspecifics.

149

4.5.3. Reproductive status and song characteristics:

As body size appears to account for a limited portion of the variation in vocalization frequencies, additional factors likely explain the remaining variation. We recorded males at different breeding stages which could have affected among-male differences in vocalization parameters. As the breeding season progresses a male’s behaviour and role typically shifts – from attracting mates and defending territories to caring for and defending young and the resources they require. Consequently, both endocrinal state and the primary function of male songs likely also undergo changes over the course of the breeding season: androgen levels would be expected to decrease (Steiger et al. 2006), while the initial function of songs as a dual signal may shift to a mainly territorial or resource-defense role later on as mate attraction and courtship should be largely confined to the early parts of the breeding season (see chapter 5).

These likely changes did not, however, appear to lead to significant changes in the fundamental song frequency characteristics that were represented by the first principal component and differed among sites (chapter 3). Pre-incubation and chick-rearing males did, however, show differences in the third principal component. While incubating males did not differ significantly from either, they did show intermediate values, possibly indicating progressive changes over the course of the breeding season. Based on the component’s loadings these changes appear to reflect a shift towards fewer, but longer introductory notes along with a decrease in the bandwidth of the terminal trill components.

A constraint between trill or note repetition rate and bandwidth has been postulated for other bird taxa where it may constitute a measure of vocal performance affecting mate choice and reproductive success (e.g. Podos 1997; Ballentine et al. 2004; Janicke et al. 2008). It is therefore very possible that such a relationship also exists for Western Sandpipers. Similarly, a rapid rendition of many short introductory notes may be a more demanding task and hence might also serve as a performance measure. Higher

150

performance songs would likely be important early in the season when song function includes mate attraction, but may be less crucial later in the season when its role shifts towards territorial defense. Hence, the observed changes may reflect a shift towards lower performance songs later in the season in line with a change in song function.

4.6. Conclusions

The fundamental frequencies of male Western Sandpiper songs and alarm calls are positively related, individually distinct and negatively correlated with body size. In the case of songs, these frequencies do not appear to be affected by a males’ reproductive stage within a season. The changes that songs did show over the course of the breeding season seem to primarily pertain to the number and duration of introductory notes as well as trill bandwidth characteristics, which possibly act as sexually selected performance measures.

Geographical variation in male song frequencies are therefore less likely to be a result of phenotypic plasticity and may be partially driven by differences in body size. A possibly stronger effect of body size on song than alarm call fundamental frequencies might also partially explain why male songs differed significantly across breeding sites while alarm calls did not (Chapter 3). Our study sheds some light onto sources of variation in Western Sandpiper songs and alarm calls. Nevertheless, we clearly know very little about what influences shorebird vocalizations and further research into the effects and communicative value of such factors as morphology, condition and behavioural contexts is needed.

4.7. References

Acuña, E., and C. Rodriguez. 2004. The Treatment of Missing Values and its Effect on Classifier Accuracy. Pages 639-647 in Classification, Clustering, and Data

151

Mining Applications (D. Banks, F. R. McMorris, P. Arabie, and W. Gaul, Eds.). Springer Berlin Heidelberg.

Araya-Ajoy, Y. M., J. Chaves-Campos, E. K. V. Kalko, and J. A. DeWoody. 2009. High-Pitched Notes during Vocal Contests Signal Genetic Diversity in Ocellated Antbirds. PLoS ONE 4:7.

Badyaev, A. V., and E. S. Leaf. 1997. Habitat associations of song characteristics in Phylloscopus and Hippolais warblers. Auk 114:40-46.

Baker, M. C. 1982. Individuality of Vocalizations in Dunlin: A Possible Acoustic Basis for Recognition of Parent by Offspring. Auk 99:771-774.

Ballentine, B., J. Hyman, and S. Nowicki. 2004. Vocal performance influences female response to male bird song: an experimental test. Behavioral Ecology 15:163-168.

Barbraud, C., A. Mariani, and P. Jouventin. 2000. Variation in call properties of the snow petrel, Pagodroma nivea, in relation to sex and body size. Australian Journal of Zoology 48:421-430.

Beckers, G. J. L., R. A. Suthers, and C. ten Cate. 2003. Mechanisms of frequency and amplitude modulation in ring dove song. Journal of Experimental Biology 206:1833- 1843.

Beecher, M. D., P. K. Stoddard, and P. Loesche. 1985. Recognition of Parent's Voices by Young Cliff Swallows. Auk 102:600-605.

Beightol, D. R., and D. E. Samuel. 1973. Sonagraphic Analysis Of American Woodcocks Peent Call. Journal of Wildlife Management 37:470-475.

152

Bennet-Clark, H. C. 1998. Size and scale effects as constraints in insect sound communication. Philosophical Transactions of the Royal Society of London Series B- Biological Sciences 353:407-419.

Berg, K. S., S. Delgado, R. Okawa, S. R. Beissinger, and J. W. Bradbury. 2011. Contact calls are used for individual mate recognition in free-ranging green-rumped parrotlets, Forpus passerinus. Animal Behaviour 81:241-248.

Bertelli, S., and P. L. Tubaro. 2002. Body mass and habitat correlates of song structure in a primitive group of birds. Biological Journal of the Linnean Society 77:423- 430.

Blumenrath, S. H., T. Dabelsteen, and S. B. Pedersen. 2007. Vocal neighbour- mate discrimination in female great tits despite high song similarity. Animal Behaviour 73:789-796.

Bolt, L. M. 2013. The Function of Howling in the Ring-Tailed Lemur (Lemur catta). International Journal of Primatology 34:157-169.

Boogert, N. J., L. A. Giraldeau, and L. Lefebvre. 2008. Song complexity correlates with learning ability in zebra finch males. Animal Behaviour 76:1735-1741.

Bowman, R. I. 1979. Adaptive Morphology of Song Dialects in Darwin's Finches. Journal Für Ornithologie 120:353-389.

Bradbury, J. W., and S. L. Vehrencamp. 2011. Principles of animal communication, 2 ed. Sinauer Associates, Inc., Sunderland, Mass.

153

Bretagnolle, V. 1996. Acoustic Communication in a Group of Nonpasserine Birds, the Petrels. Pages 160-177 in Ecology and evolution of acoustic communication in birds (D. Kroodsma, and E. H. Miller, Eds.). Comstock/Cornell University Press, Ithaca.

Bried, J., and P. Jouventin. 1997. Morphological and vocal variation among subspecies of the black-faced Sheathbill. Condor 99:818-825.

Brown, C. and D. Ward 1990. The morphology of the syrinx in the Charadriiformes (Aves): possible phylogenetic implications. Bonner Zoologische Beiträge 41:95-107.

Brown, R. G. B. 1962. The Aggressive and Distraction Behaviour of the Western Sandpiper Ereunetes mauri. Ibis 104:1-12.

Budka, M., and T. S. Osiejuk. 2013. Neighbour-stranger call discrimination in a nocturnal rail species, the Corncrake Crex crex. Journal of Ornithology 154:685-694.

Cardoso, G. C., A. T. Mamede, J. W. Atwell, P. G. Mota, E. D. Ketterson, and T. D. Price. 2008. Song Frequency Does Not Reflect Differences in Body Size among Males in Two Oscine Species. Ethology 114:1084-1093.

Cardoso, G. C., and P. G. Mota. 2007. Song diversification and complexity in canaries and seedeaters (Serinus spp.). Biological Journal of the Linnean Society 92:183- 194.

Castellano, S., and M. Gamba. 2011. Marathon callers: acoustic variation during sustained calling in treefrogs. Ethology Ecology & Evolution 23:329-342.

Castellano, S., and A. Rosso. 2007. Female preferences for multiple attributes in the acoustic signals of the Italian treefrog, Hyla intermedia. Behavioral Ecology and Sociobiology 61:1293-1302.

154

Catchpole, C. K. 1983. Variation in the Song of the Great Reed Warbler Acrocephalus arundinaceus in Relation to Mate Attraction and Territorial Defense. Animal Behaviour 31:1217-1225.

Catchpole, C. K. 1987. Bird Song, Sexual Selection And Female Choice. Trends in Ecology & Evolution 2:94-97.

Catchpole, C. K., and P. J. B. Slater. 2008. Bird song: biological themes and variations. Cambridge University Press, Cambridge, UK.

Charlton, B. D., W. A. H. Ellis, J. Brumm, K. Nilsson, and W. T. Fitch. 2012a. Female koalas prefer bellows in which lower formants indicate larger males. Animal Behaviour 84:1565-1571.

Charlton, B. D., R. R. Swaisgood, Z. H. Zhihe, and R. J. Snyder. 2012b. Giant pandas attend to androgen-related variation in male bleats. Behavioral Ecology and Sociobiology 66:969-974.

Charrier, I., N. Mathevon, and P. Jouventin. 2003. Vocal signature recognition of mothers by fur seal pups. Animal Behaviour 65:543-550.

Charrier, I., B. J. Pitcher, and R. G. Harcourt. 2009. Vocal recognition of mothers by Australian sea lion pups: individual signature and environmental constraints. Animal Behaviour 78:1127-1134.

Clark, J. A., P. D. Boersma, and D. M. Olmsted. 2006. Name that tune: call discrimination and individual recognition in Magellanic penguins. Animal Behaviour 72:1141-1148.

155

Crawford, J. D., A. P. Cook, and A. S. Heberlein. 1997. Bioacoustic behavior of African fishes (Mormyridae): Potential cues for species and individual recognition in Pollimyrus. Journal of the Acoustical Society of America 102:1200-1212.

Dale, J., D. B. Lank, and H. K. Reeve. 2001. Signaling individual identity versus quality: A model and case studies with ruffs, queleas, and house finches. American Naturalist 158:75-86.

Darden, S. K., T. Dabelsteen, and S. B. Pedersen. 2003. A potential tool for swift fox (Vulpes velox) conservation: Individuality of long-range barking sequences. Journal of Mammalogy 84:1417-1427.

Darwin, C. 1871. The Descent of Man, and Selection in Relation to Sex. John Murray, London.

Dentressangle, F., T. Aubin, and N. Mathevon. 2012. Males use time whereas females prefer harmony: individual call recognition in the dimorphic blue-footed booby. Animal Behaviour 84:413-420.

Digweed, S. M., D. Rendall, and T. Imbeau. 2012. Who's your neighbor? Acoustic cues to individual identity in red squirrel Tamiasciurus hudsonicus rattle calls. Current Zoology 58:758-764.

Douglas, H. D. I., and W. E. Conner. 1999. Is there a sound reception window in coastal environments? Evidence from shorebird communication systems. Naturwissenschaften 86:228-230.

156

Draganoiu, T. I., A. Moreau, L. Ravaux, W. Bonckaert, and N. Mathevon. 2014. Song stability and neighbour recognition in a migratory songbird, the black redstart. Behaviour 151:435-453.

Dragonetti, M. 2007. Individuality in Scops Owl Otus scops vocalisations. Bioacoustics - the International Journal of Animal Sound and Its Recording 16:147-172.

Dragonetti, M., C. Caccamo, E. Pollonara, N. E. Baldaccini, and D. Giunchi. 2013. A weak individual signature might not allow chick call recognition by parent Stone Curlews Burhinus oedicnemus. Bioacoustics - the International Journal of Animal Sound and Its Recording 22:17-32.

Duellman, W. E., and R. A. Pyles. 1983. Acoustic Resource Partitioning in Anuran Communities. Copeia:639-649.

Eckenweber, M., and M. Knörnschild. 2013. Social influences on territorial signaling in male greater sac-winged bats. Behavioral Ecology and Sociobiology 67:639- 648.

Falls, J. B. 1982. Individual recognition by sound in birds. Pages 237-278 in Acoustic Communication in Birds, vol. 2: Song Learning and its Consequences (D. E. Kroodsma, and E. H. Miller, Eds.). Academic Press, New York.

Farmer, A., R. T. Holmes, and F. A. Pitelka. 2013. Pectoral Sandpiper (Calidris melanotos), The Birds of North America Online (A. Poole, Ed.). Cornell Lab of Ornithology, Ithaca. http://bna.birds.cornell.edu/bna.

Feng, A. S., V. S. Arch, Z. L. Yu, X. J. Yu, Z. M. Xu, and J. X. Shen. 2009. Neighbor-Stranger Discrimination in Concave-Eared Torrent Frogs, Odorrana tormota. Ethology 115:851-856.

157

Fitch, W. T., and M. D. Hauser. 2003. Unpacking Honesty: Vertebrate Vocal Production and the Evolution of Acoustic Signals. in Acoustic Communication. (A. M. Simmons, R. R. Fay, and A. N. Popper, Eds.). Springer Verlag, Berlin.

Fletcher, N. H. 2004. A simple frequency-scaling rule for animal communication. Journal of the Acoustical Society of America 115:2334-2338.

Foote, J. R., E. Palazzi, and D. J. Mennill. 2013. Songs of the Eastern Phoebe, a suboscine songbird, are individually distinctive but do not vary geographically. Bioacoustics - the International Journal of Animal Sound and Its Recording 22:137-151.

Forstmeier, W., C. Burger, K. Temnow, and S. Deregnaucourt. 2009. The Genetic Basis of Zebra Finch Vocalizations. Evolution 63:2114-2130.

Galeotti, P., N. Saino, R. Sacchi, and A. P. Moller. 1997. Song correlates with social context, testosterone and body condition in male barn swallows. Animal Behaviour 53:687-700.

Geberzahn, N., W. Goymann, C. Muck, and C. ten Cate. 2009. Females alter their song when challenged in a sex-role reversed bird species. Behavioral Ecology and Sociobiology 64:193-204.

Gerhardt, H. C. 1992. Multiple messages in acoustic signals. Seminars in the Neurosciences 4:391-400.

Giacoma, C., C. Zugolaro, and L. Beani. 1997. The advertisement calls of the green toad (Bufo viridis): Variability and role in mate choice. Herpetologica 53:454-464.

Hall, M. L., S. A. Kingma, and A. Peters. 2013. Male Songbird Indicates Body Size with Low-Pitched Advertising Songs. PLoS ONE 8:5.

158

Handford, P., and S. C. Lougheed. 1991. Variation in Duration and Frequency Characters in the Song of the Rufous-Collared Sparrow, Zonotrichia capensis, with Respect to Habitat, Trill Dialects and Body Size. Condor 93:644-658.

Hardouin, L. A., P. Tabel, and V. Bretagnolle. 2006. Neighbour-stranger discrimination in the little owl, Athene noctua. Animal Behaviour 72:105-112.

Hill, B. G., and M. R. Lein. 1987. Function of Frequency-Shifted Songs of Black- Capped Chickadees. Condor 89:914-915.

Holmes, R. T. 1971. Density, Habitat and the Mating System of the Western Sandpiper (Calidris mauri). Oecologia 7:191-208.

Holmes, R. T. 1972. Ecological Factors Influencing the Breeding Season Schedule of Western Sandpipers (Calidris mauri) in Subarctic Alaska. American Midland Naturalist 87:472-491.

Holmes, R. T. 1973. Social Behavior of Breeding Western Sandpipers Calidris mauri. Ibis 115:107-123.

Howe, M. A. 1972. Pair bond formation and maintenance in Wilson's Phalarope, Phalaropus tricolor Ph.D. Thesis, University of Minnesota, Minneapolis.

Janicke, T., S. Hahn, M. S. Ritz, and H. U. Peter. 2008. Vocal performance reflects individual quality in a nonpasserine. Animal Behaviour 75:91-98.

159

Johnson, J. M. 2006. Demography and Behavior of Western Sandpipers (Calidris mauri) Breeding on the Yukon-Kuskokwim River Delta, Alaska. Ph.D., Virginia Polytechnic Institute and State University, Blacksburg, Virginia.

Johnson, M., S. Aref, and J. R. Walters. 2008. Parent-offspring communication in the Western Sandpiper. Behavioral Ecology 19:489-501.

Johnson, M., and B. McCaffery. 2004. Use of upland tundra habitats by Western and Rock Sandpipers during brood-rearing on the Yukon-Kuskokwim Delta, Alaska. Wader Study Group Bull. 103:36-39.

Jurisevic, M. A., and K. J. Sanderson. 1998. A comparative analysis of distress call structure in Australian passerine and non-passerine species: influence of size and phylogeny. Journal of Avian Biology 29:61-71.

Keen, S. C., C. D. Meliza, and D. R. Rubenstein. 2013. Flight calls signal group and individual identity but not kinship in a cooperatively breeding bird. Behavioral Ecology 24:1279-1285.

Knörnschild, M., M. Feifel, and E. K. V. Kalko. 2013. Mother-offspring recognition in the bat Carollia perspicillata. Animal Behaviour 86:941-948.

Konishi, M. 1970. Comparative Neurophysiological Studies of Hearing and Vocalizations in Songbirds. Zeitschrift Für Vergleichende Physiologie 66:257-272.

Koren, L., and E. Geffen. 2009. Complex call in male rock hyrax (Procavia capensis): a multi-information distributing channel. Behavioral Ecology and Sociobiology 63:581-590.

160

Koren, L., and E. Geffen. 2011. Individual identity is communicated through multiple pathways in male rock hyrax (Procavia capensis) songs. Behavioral Ecology and Sociobiology 65:675-684.

Laiolo, P., and A. Rolando. 2003. The evolution of vocalisations in the genus Corvus: effects of phylogeny, morphology and habitat. Evolutionary Ecology 17:111- 123.

Laiolo, P., A. Rolando, A. Delestrade, and A. De Sanctis. 2004. Vocalizations and morphology: interpreting the divergence among populations of Chough Pyrrhocorax pyrrhocorax and Alpine Chough P. graculus. Bird Study 51:248-255.

Lambrechts, M. M. 1996. Organization of birdsong and constraints on performance. Pages 305-320 in Ecology and evolution of acoustic communication in birds (D. E. Kroodsma, and E. H. Miller, Eds.). Comstock/Cornell University Press, Ithaca, NY.

Lampe, H. M., and Y. O. Espmark. 1994. Song Structure Reflects Male Quality In Pied Flycatchers, Ficedula hypoleuca. Animal Behaviour 47:869-876.

Lovell, S. F., and M. R. Lein. 2004a. Neighbor-stranger discrimination by song in a suboscine bird, the alder flycatcher, Empidonax alnorum. Behavioral Ecology 15:799- 804.

Lovell, S. F., and M. R. Lein. 2004b. Song variation in a population of Alder Flycatchers. Journal of Field Ornithology 75:146-151.

161

Mager, J. N., C. Walcott, and W. H. Piper. 2010. Common Loons can differentiate yodels of neighboring and non-neighboring conspecifics. Journal of Field Ornithology 81:392-401.

Mahler, B., and D. Gil. 2009. The Evolution of Song in the Phylloscopus Leaf Warblers (Aves: Sylviidae): A Tale of Sexual Selection, Habitat Adaptation, and Morphological Constraints. Pages 35-66 in Advances in the Study of Behavior, Vol 40, vol. 40 (M. Naguib, K. Zuberbuhler, N. S. Clayton, and V. M. Janik, Eds.). Elsevier Academic Press Inc, San Diego.

Marler, P. 2004. Bird calls - Their potential for behavioral neurobiology. Behavioral Neurobiology of Birdsong 1016:31-44.

Martin, J. P., S. M. Doucet, R. C. Knox, and D. J. Mennill. 2011. Body size correlates negatively with the frequency of distress calls and songs of Neotropical birds. Journal of Field Ornithology 82:259-268.

Masters, W. M., K. A. S. Raver, and K. A. Kazial. 1995. Sonar signals of big brown bats, Eptesicus fuscus, contain information about individual identity, age and family affiliation. Animal Behaviour 50:1243-1260.

Matthews, J. N., L. E. Rendell, J. C. D. Gordon, and D. W. MacDonald. 1999. A review of frequency and time parameters of cetacean tonal calls. . Bioacoustics - the International Journal of Animal Sound and Its Recording 10:47—71.

McGregor, P. K. 1993. Signaling in Territorial Systems: A Context for Individual Identification, Ranging and Eavesdropping. Philosophical Transactions of the Royal Society of London Series B-Biological Sciences 340:237-244.

162

Melendez, K. V., and A. S. Feng. 2010. Communication calls of little brown bats display individual-specific characteristics. Journal of the Acoustical Society of America 128:919-923.

Miller, E. H. 1984. Communication in breeding shorebirds. Pages 169-241 in Shorebirds: Breeding Behavior and Populations (J. Burger, Olla, B.L., Eds.). Plenum Press, New York.

Miller, E. H. 1986. Components of Variation in Nuptial Calls of the Least Sandpiper (Calidris minutilla; Aves, Scolopacidae). Systematic Zoology 35:400-413.

Miller, E. H., W. W. H. Gunn, and B. N. Veprintsev. 1988. Breeding Vocalizations of Bairds Sandpiper Calidris bairdii and Related Species, with Remarks on Phylogeny and Adaptation. Ornis Scandinavica 19:257-267.

Morton, E. S. 1977. Occurrence and Significance of Motivation-Structural Rules in Some Bird and Mammal Sounds. American Naturalist 111:855-869.

Naguib, M., K. Hammerschmidt, and J. Wirth. 2001. Microgeographic variation, habitat effects and individual signature cues in calls of chiffchaffs Phylloscopus collybita canarensis. Ethology 107:341-355.

163

Nelson, D. A., and A. Poesel. 2007. Segregation of information in a complex acoustic signal: individual and dialect identity in white-crowned sparrow song. Animal Behaviour 74:1073-1084.

Nolan, P. M., and G. E. Hill. 2004. Female choice for song characteristics in the house finch. Animal Behaviour 67:403-410.

Nowicki, S., S. Peters, and J. Podos. 1998. Song learning, early nutrition and sexual selection in songbirds. American Zoologist 38:179-190.

Nowicki, S., and W. A. Searcy. 2004. Song function and the evolution of female preferences - Why birds sing, why brains matter. Annals of the New York Academy of Sciences 1016:704-723.

Nowicki, S., W. A. Searcy, and S. Peters. 2002. Brain development, song learning and mate choice in birds: a review and experimental test of the "nutritional stress hypothesis". Journal of Comparative Physiology a-Neuroethology Sensory Neural and Behavioral Physiology 188:1003-1014.

Odom, K. J., and D. J. Mennill. 2012. Inconsistent Geographic Variation in the Calls and Duets of Barred Owls (Strix varia) across an Area of Genetic Introgression. Auk 129:387-398.

Odom, K. J., J. C. Slaght, and R. J. Gutierrez. 2013. Distinctiveness in the Territorial Calls of Great Horned Owls within and among Years. Journal of Raptor Research 47:21-30.

Osiejuk, T. S., K. Ratynska, J. P. Cygan, and S. Dale. 2005. Frequency shift in homologue syllables of the Ortolan Bunting Emberiza hortulana. Behavioural Processes 68:69-83.

164

Patel, R., R. A. Mulder, and G. C. Cardoso. 2010. What Makes Vocalisation Frequency an Unreliable Signal of Body Size in Birds? A Study on Black Swans. Ethology 116:554-563.

Pettitt, B. A., G. R. Bourne, and M. A. Bee. 2013. Advertisement Call Variation in the Golden Rocket Frog (Anomaloglossus beebei): Evidence for Individual Distinctiveness. Ethology 119:244-256.

Pimentel, R. A., and D. F. Frey. 1978. Multivariate Analysis of Variance and Discriminant Analysis. Pages 247-274 in Quantitative Ethology. (P. W. Colgan, Ed.). Wiley, New York.

Podos, J. 1997. A performance constraint on the evolution of trilled vocalizations in a songbird family (Passeriformes:Emberizidae). Evolution 51:537-551.

Podos, J., S. K. Huber, and B. Taft. 2004. Bird song: The interface of evolution and mechanism. Annual Review of Ecology Evolution and Systematics 35:55-87.

Policht, R., M. Petru, L. Lastimoza, and L. Suarez. 2009. Potential for the use of vocal individuality as a conservation research tool in two threatened Philippine hornbill species, the Visayan Hornbill and the Rufous-headed Hornbill. Bird Conservation International 19:83-97.

Price, T. 2007. Speciation in Birds. Roberts and Company Publishers, Greenwood Village, Colorado.

Rek, P., and T. S. Osiejuk. 2010. Sophistication and simplicity: conventional communication in a rudimentary system. Behavioral Ecology 21:1203-1210.

Rios-Chelen, A. A., C. M. Garcia, and K. Riebel. 2005. Variation in the song of a sub-oscine, the vermilion flycatcher. Behaviour 142:1115-1132.

165

Rising, J. D., and K. M. Somers. 1989. The Measurement of Overall Body Size in Birds. Auk 106:666-674.

Ritchie, M. G., E. J. Halsey, and J. M. Gleason. 1999. Drosophila song as a species-specific mating signal and the behavioural importance of Kyriacou & Hall cycles in D. melanogaster song. Animal Behaviour 58:649-657.

Robisson, P., T. Aubin, and J. C. Bremond. 1993. Individuality in the Voice of the Emperor Penguin Aptenodytes forsteri - Adaptation to a Noisy Environment. Ethology 94:279-290.

Ruthrauff, D. R., and B. J. McCaffery. 2005. Survival of Western Sandpiper broods on the Yukon-Kuskokwim Delta, Alaska. Condor 107:597-604.

Ryan, M. J., and E. A. Brenowitz. 1985. The Role of Body Size, Phylogeny, and Ambient Noise in the Evolution of Bird Song. American Naturalist 126:87-100.

Saarikettu, M., J. O. Liimatainen, and A. Hoikkala. 2005. The role of male courtship song in species recognition in Drosophila montana. Behavior Genetics 35:257- 263.

Schubert, M. 1976. Über die Variabilität von Lockrufen des Gimpels Pyrrhula pyrrhula. Ardea 64:62-71.

Searby, A., P. Jouventin, and T. Aubin. 2004. Acoustic recognition in macaroni penguins: an original signature system. Animal Behaviour 67:615-625.

Searcy, W. A., and M. Andersson. 1986. Sexual Selection And The Evolution Of Song. Annual Review of Ecology and Systematics 17:507-533.

166

Seddon, N. 2005. Ecological adaptation and species recognition drives vocal evolution in neotropical suboscine birds. Evolution 59:200-215.

Shy, E. 1983. The Relation of Geographical Variation in Song to Habitat Characteristics and Body Size in North-American Tanagers (Thraupinae, Piranga). Behavioral Ecology and Sociobiology 12:71-76.

Skierczyński, M., K. M. Czarnecka, and T. S. Osiejuk. 2007. Neighbour-stranger song discrimination in territorial ortolan bunting Emberiza hortulana males. Journal of Avian Biology 38:415-420.

Spehar, S. N., and A. Di Fiore. 2013. Loud calls as a mechanism of social coordination in a fission-fusion taxon, the white-bellied spider monkey (Ateles belzebuth). Behavioral Ecology and Sociobiology 67:947-961.

Steiger, S. S., W. Goymann, and B. Kempenaers. 2006. Plasma steroid hormones in two Arctic-breeding shorebirds: Monogamy versus polygyny. General and Comparative Endocrinology 147:133-140.

Stoddard, P. 1996. Vocal recognition of neighbors by territorial passerines. Pages 356-374 in Ecology and evolution of acoustic communication in birds (D. E. Kroodsma, and E. H. Miller, Eds.). Comstock/Cornell University Press, Ithaca.

Stoddard, P. K., and M. D. Beecher. 1983. Parental Recognition of Offspring in the Cliff Swallow. Auk 100:795-799.

Stoffberg, S., D. S. Jacobs, and C. A. Matthee. 2011. The Divergence of Echolocation Frequency in Horseshoe Bats: Moth Hearing, Body Size or Habitat? Journal of Mammalian Evolution 18:117-129.

167

Sung, H. C., and E. H. Miller. 2007. Variation and individuality in display-flight calls of male piping plovers. Condor 109:468-475.

Terry, A. M., T. M. Peake, and P. K. McGregor. 2005. The role of vocal individuality in conservation. Frontiers in Zoology 2.

Theis, K. R., K. M. Greene, S. R. Benson-Amram, and K. F. Holekamp. 2007. Sources of variation in the long-distance vocalizations of spotted hyenas. Behaviour 144:557-584.

Thom, M. D. F., and C. Dytham. 2012. Female Choosiness Leads to the Evolution of Individually Distinctive Males. Evolution 66:3736-3742.

Tibbetts, E. A., and J. Dale. 2007. Individual recognition: it is good to be different. Trends in Ecology & Evolution 22:529-537.

Tripovich, J. S., I. Charrier, T. L. Rogers, R. Canfield, and J. P. Y. Arnould. 2008. Who goes there? Differential responses to neighbor and stranger vocalizations in male Australian fur seals. Marine Mammal Science 24:941-948.

Tubaro, P. L., and B. Mahler. 1998. Acoustic frequencies and body mass in new world doves. Condor 100:54-61.

Tyack, P. L. 2008. Convergence of calls as animals form social bonds, active compensation for noisy communication channels, and the evolution of vocal learning in mammals. Journal of Comparative Psychology 122:319-331.

Vannoni, E., and A. G. McElligott. 2009. Fallow bucks get hoarse: vocal fatigue as a possible signal to conspecifics. Animal Behaviour 78:3-10.

168

Vaughan, N., G. Jones, and S. Harris. 1997. Identification of British bat species by multivariate analysis of echolocation call parameters. Bioacoustics - the International Journal of Animal Sound and Its Recording 7:189-207.

Vehrencamp, S. L., J. Yantachka, M. L. Hall, and S. R. de Kort. 2013. Trill performance components vary with age, season, and motivation in the banded wren. Behavioral Ecology and Sociobiology 67:409-419.

Vignal, C., N. Mathevon, and S. Mottin. 2004. Audience drives male songbird response to partner's voice. Nature 430:448-451.

von Frisch, O. 1956. Zur Brutbiologie und Jugendentwicklung des Brachvogels (Numenius arquata L.). Z. Tierpsychol. 13:50-81.

von Frisch, O. 1959. Zur Jugendentwicklung, Brutbiologie und vergleichenden Ethologie der Limicolen. Z. Tierpsychol. 16:545-583.

Wallschläger, D. 1980. Correlation of Song Frequency and Body-Weight in Passerine Birds. Experientia 36:412-412.

Wei, M., H. Lloyd, and Y. Y. Zhang. 2011. Neighbour-stranger discrimination by Yellow-bellied Tit Parus venustulus: evidence for the "dear-enemy" effect. Journal of Ornithology 152:431-438.

Whitfield, D. P. 1986. Plumage variability and territoriality in breeding Turnstone Arenaria interpres: status signaling or individual recognition. Animal Behaviour 34:1471-1482.

Whitfield, D. P. 1988. The social significance of plumage variability in wintering turnstone Arenaria interpres. Animal Behaviour 36:408-415.

169

Wiley, R. H. 1991. Associations of Song Properties with Habitats for Territorial Oscine Birds of Eastern North America. American Naturalist 138:973-993.

Wiley, R. H. 2005. Individuality in songs of Acadian flycatchers and recognition of neighbours. Animal Behaviour 70:237-247.

Wilson, D. R., and D. J. Mennill. 2010. Black-capped chickadees, Poecile atricapillus, can use individually distinctive songs to discriminate among conspecifics. Animal Behaviour 79:1267-1275.

Ydenberg, R. C., L. A. Giraldeau, and J. B. Falls. 1988. Neighbors, strangers, and the asymmetric war of attrition. Animal Behaviour 36:343-347.

Young, F. W. 2003. ViSta “The Visual Statistics System”. [computer software] [on-line] Available: URL: http://forrest.psych.unc.edu/research/index.html.

Zhang, B., J. K. Cheng, Y. Han, L. Q. Ji, and F. M. Shi. 2012. An acoustic system for the individual recognition of insects. Journal of the Acoustical Society of America 131:2859-2865.

170

4.8. Figures 171

a. b.

Figure 4-1. Scatterplot of discriminant function 1 (DF1) and 2 (DF2) values for male songs (a) and alarm calls (b). Each datapoint in the scatterplot represents a single song (a) or alarm call (b). Individual males (n=13 (a) and n=7 (b)) are represented by different symbols, their centroids through numbered squares. Frequency and temporal variables both entered into a Principal Component Analysis. We included all PCs with eigenvalues > 1 in a stepwise DFA model (threshold of F=3.84 for entry and F=2.71 for removal of a variable; separate groups covariance matrix; prior probabilities based on sample size). For songs (a) the DFA assigned 92.8% of songs to the correct males based on the first seven discriminant functions, with DF 1 and 2 accounting for 60.2 % of the variation. All alarm calls (b) were correctly assigned using the first five discriminant functions, DF1 and 2 accounted for 77.4% of the variation.

r2=0.345

Figure 4-2. Correlation of alarm call center frequency with song frequency PC1: Each datapoint represents one male (n=21) and data from all four breeding sites were entered into the analysis. The variation in alarm call frequency explained 34.5% of the variation in the first song frequency Principal Component (P=0.003), revealing a strongly positive relationship between the frequencies of the two vocalization types.

172

r2=0.103 173

r2=0.138

Figure 4-4. Effect of reproductive state on song characteristics. We conducted a PCA on the song parameters of pre-incubation, incubating and brood-rearing males at the Nome breeding site. While reproductive stage did not appear to affect PC1 and 2 (not shown), PC3 differed among pre-incubation and chick-rearing males (n=30, F=5.5, df=2 and 27, P=0.01). PC3 accounted for 10.3% of the total variation in song characteristics and primarily represented the number of introductory notes, their duration (inverse loadings) and bandwidth characteristics of terminal trill components. Error bars represent 95% confidence intervals.

174

4.9. Appendix

Table 4-1. Component loadings for the Principal Components (PCs) used in Discriminant Function Analyses (DFA) of songs (a.) and alarm calls (b.). Only PCs with eigenvalues > 1 entered the DFA and are listed here. a. SONGS Component Loadings Selection and variable PC1 PC2 PC3 PC4 PC5 PC6 PC7 PC8 PC9 PC10 i1 Peak Frequency (Hz) 0.373 0.083 -0.111 0.089 -0.46 0.302 -0.524 -0.128 0.047 0.132 i2 Peak Frequency (Hz) 0.605 0.371 -0.031 -0.231 0.166 -0.02 0.062 0.178 0.112 -0.252 i3 Peak Frequency (Hz) 0.701 0.284 -0.125 -0.067 0.073 0.12 0 0.29 0.124 -0.033 i4 Peak Frequency (Hz) 0.699 0.329 -0.107 0.08 -0.013 0.096 0.008 0.204 0.173 -0.167 0.812 0.435 -0.036 0.108 -0.027 0.003 0.122 -0.032 0.092 0.033 175 t1 Peak Frequency (Hz) t2 Peak Frequency (Hz) 0.758 0.444 -0.085 0.214 -0.104 -0.003 0.145 -0.125 -0.054 0.043 t3 Peak Frequency (Hz) 0.709 0.385 -0.107 0.311 -0.135 -0.045 0.145 -0.086 0.029 0.165 t4 Peak Frequency (Hz) 0.759 -0.448 0.095 -0.111 0.026 -0.011 -0.011 -0.132 -0.028 0.127 t5 Peak Frequency (Hz) 0.742 -0.504 0.122 -0.185 0.019 0.068 -0.019 -0.138 0.062 0.014 t6 Peak Frequency (Hz) 0.649 -0.492 0.174 -0.134 -0.024 0.021 0.006 -0.232 0.123 -0.228 t7 Peak Frequency (Hz) 0.724 -0.308 0.143 -0.042 0.194 -0.159 -0.043 0.064 -0.267 0.079 i1 Center Frequency (Hz) 0.374 0.083 -0.052 0.038 -0.426 0.404 -0.568 -0.095 0.132 0.206 i2 Center Frequency (Hz) 0.558 0.579 0.135 -0.299 0.05 0.104 -0.08 0.099 -0.168 -0.198 i3 Center Frequency (Hz) 0.683 0.542 0.067 -0.212 -0.024 0.09 -0.03 0.109 -0.11 -0.162 i4 Center Frequency (Hz) 0.717 0.574 -0.021 -0.042 -0.076 0.043 0.052 0.003 -0.016 -0.072 t1 Center Frequency (Hz) 0.796 0.441 -0.142 0.052 0.01 -0.035 0.076 -0.013 0.137 0.007 t2 Center Frequency (Hz) 0.753 0.486 -0.24 0.168 -0.04 -0.076 0.135 -0.089 -0.019 0.058 t3 Center Frequency (Hz) 0.678 0.356 -0.344 0.176 0.013 -0.083 0.096 -0.162 0.013 0.218

SONGS Component Loadings Selection and variable PC1 PC2 PC3 PC4 PC5 PC6 PC7 PC8 PC9 PC10 t4 Center Frequency (Hz) 0.762 -0.493 0.081 -0.168 0.047 0 0.005 -0.126 0.007 0.12 t5 Center Frequency (Hz) 0.757 -0.527 0.112 -0.182 0.053 0.044 -0.035 -0.118 0.055 0.023 t6 Center Frequency (Hz) 0.65 -0.533 0.159 -0.144 -0.013 0.019 0.048 -0.216 0.118 -0.218 t7 Center Frequency (Hz) 0.757 -0.354 0.169 -0.059 0.277 -0.083 -0.072 0.013 -0.203 0.123 i1 90% Bandwidth (Hz) -0.213 0.136 0.234 0.058 0.106 -0.079 -0.072 -0.452 0.385 0.441 i2 90% Bandwidth (Hz) -0.088 -0.036 0.438 -0.042 -0.141 -0.265 0.055 0.055 0.587 -0.158 i3 90% Bandwidth (Hz) -0.101 -0.039 0.651 -0.179 -0.346 0.155 0.124 0.211 0.323 -0.018 i4 90% Bandwidth (Hz) -0.008 -0.055 0.725 -0.073 -0.307 0.199 0.184 0.104 0.115 -0.045 t1 90% Bandwidth (Hz) 0.1 0.106 0.629 0.248 -0.324 0.209 0.293 0.051 -0.141 0.17 t2 90% Bandwidth (Hz) 0.048 0.03 0.756 0.222 -0.319 0.252 0.235 0.016 -0.162 0.123 t3 90% Bandwidth (Hz) -0.067 0.039 0.768 0.25 -0.044 0.142 0.139 0.061 -0.287 0.046 176 t4 90% Bandwidth (Hz) 0.178 0.091 0.458 0.44 0.481 -0.094 -0.269 0.046 -0.117 -0.006 t5 90% Bandwidth (Hz) 0.14 0.111 0.419 0.441 0.355 -0.153 -0.351 0.025 0.057 -0.117 t6 90% Bandwidth (Hz) 0.11 -0.046 0.363 0.381 0.225 -0.237 -0.443 -0.016 0.238 -0.19 t7 90% Bandwidth (Hz) 0.239 0 0.504 0.471 0.295 0.069 -0.085 0.094 -0.076 -0.086 i1-t1 90% Bandwidth (Hz) -0.018 0.274 0.173 0.077 0.427 -0.464 0.542 -0.182 0.153 0.182 t1-t6 90% Bandwidth (Hz) -0.29 0.667 0.066 0.34 0.06 -0.139 -0.272 0.198 0.024 0.124 i1 Duration (ms) 0.115 -0.127 -0.301 -0.001 -0.33 -0.027 0.136 0.357 0.001 0.08 i2 Duration (ms) 0.026 -0.461 -0.368 0.442 0.202 0.423 0.236 0.047 0.141 -0.064 i3 Duration (ms) 0.137 -0.473 -0.373 0.521 0.221 0.41 0.187 0.047 0.094 -0.027 i4 Duration (ms) 0.068 -0.397 -0.27 0.43 0.355 0.491 0.051 0.108 0.094 0.031 t2 Duration (ms) 0.149 -0.246 -0.059 -0.26 0.305 0.002 -0.081 0.557 0.169 0.365 t3 Duration (ms) 0.174 -0.085 0.165 -0.51 0.381 0.023 -0.166 0.355 -0.011 0.305 t7 Duration (ms) 0.52 -0.288 0.23 -0.127 0.052 -0.099 0.088 0.238 0.085 0.126 Song Duration (ms) -0.321 0.479 0.105 -0.285 0.472 0.493 0.098 -0.128 0.1 -0.006

SONGS Component Loadings Selection and variable PC1 PC2 PC3 PC4 PC5 PC6 PC7 PC8 PC9 PC10 Intro Duration (ms) -0.205 0.299 0.161 -0.36 0.544 0.503 0.092 -0.211 0.021 -0.051 Trill Duration (ms) -0.404 0.58 -0.072 0.044 0.041 0.203 0.071 0.123 0.234 0.078 Intro Number of notes -0.271 0.526 0.288 -0.444 0.241 0.213 -0.152 -0.31 -0.087 0.052

b. ALARM CALLS Component Loadings Selection and variable PC1 PC2 PC3 PC4 PC5 PC6 1 Peak Frequency (Hz) 0.831 -0.078 0.033 0.104 0.182 0.039 2 Peak Frequency (Hz) 0.434 0.414 0.303 -0.535 -0.392 0.159 3 Peak Frequency (Hz) 0.697 0.154 0.178 0.092 -0.056 -0.379 4 Peak Frequency (Hz) 0.589 -0.533 0.112 0.173 0.052 0.111 177 1 Center Frequency (Hz) 0.842 -0.118 0.162 0.31 -0.083 0.082 2 Center Frequency (Hz) 0.521 0.47 0.405 -0.369 -0.264 0.245 3 Center Frequency (Hz) 0.759 0.127 0.4 0.185 -0.157 -0.189 4 Center Frequency (Hz) 0.707 -0.385 0.024 0.314 -0.111 0.219 1 90% Bandwidth (Hz) 0.35 0.663 -0.479 0.193 0.053 0.158 2 90% Bandwidth (Hz) 0.216 0.627 -0.398 -0.344 -0.036 0.046 3 90% Bandwidth (Hz) 0.273 0.756 -0.405 0.274 0.112 -0.068 4 90% Bandwidth (Hz) 0.046 -0.33 -0.245 -0.03 0.001 0.781 1 Duration (ms) -0.089 0.478 0.559 0.149 0.6 0.203 2 Duration (ms) -0.084 -0.308 0.36 -0.591 0.285 -0.06 3 Duration (ms) 0.623 -0.094 -0.289 -0.437 0.471 -0.019 4 Duration (ms) 0.679 -0.175 -0.181 -0.199 0.436 -0.114 1 Number of notes -0.48 0.507 0.539 0.308 0.253 0.19

Table 4-2. Contributions of the Principal Component variables (PCs) to each Discriminant Function for songs (a.) and alarm calls (b.). Only PCs that entered the analysis are shown and are ordered by the absolute size of their correlation within the function.

Function SONGS 1 2 3 4 5 6 7 PC4 -0.581 -0.226 -0.197 0.369 -0.579 0.220 0.228 PC1 -0.271 0.637 -0.130 0.380 0.599 0.027 -0.024 PC3 0.350 0.089 -0.530 0.527 -0.285 -0.402 0.260 PC5 0.178 0.036 0.434 0.582 -0.165 0.470 -0.438 PC2 0.126 -0.103 -0.314 -0.098 0.265 0.807 0.378 PC8 0.064 0.393 0.336 -0.122 -0.390 0.089 0.744 PC6 -0.019 -0.369 0.302 0.340 0.482 -0.259 0.598

ALARM Function CALLS 1 2 3 4 5 PC3 -0.279 0.349 0.739 0.493 -0.103 PC2 0.152 0.625 -0.002 -0.751 -0.144 PC4 -0.032 0.317 -0.485 0.691 -0.431 PC5 0.205 0.173 -0.007 0.353 0.896 PC1 0.575 -0.115 0.355 0.340 -0.644

178

Table 4-3. Component loadings for the first Principal Component (song frequency PC1) of the song frequency Principal Component Analysis for males in the alarm call (a.) and morphology (b.) correlation analyses. Following a parallel analysis, solely the first Principal Component was retained for further analysis; variables are listed in order of the highest loadings onto PC1. For details on selections and variables see chapter 3. a. b. Selection and variable PC1 Selection and variable PC1 i4 Center Frequency (Hz) 0.939 t2 Center Frequency (Hz) 0.947 t2 Peak Frequency (Hz) 0.93 i4 Center Frequency (Hz) 0.944 t2 Center Frequency (Hz) 0.927 t2 Peak Frequency (Hz) 0.944 i4 Peak Frequency (Hz) 0.918 i4 Peak Frequency (Hz) 0.921 t3 Center Frequency (Hz) 0.877 i3 Peak Frequency (Hz) 0.886 t3 Peak Frequency (Hz) 0.873 t3 Center Frequency (Hz) 0.878 i3 Peak Frequency (Hz) 0.866 t3 Peak Frequency (Hz) 0.867 i3 Center Frequency (Hz) 0.839 i3 Center Frequency (Hz) 0.854 i1 Center Frequency (Hz) 0.818 i1 Center Frequency (Hz) 0.8 i1 Peak Frequency (Hz) 0.714 i2 Peak Frequency (Hz) 0.719 i2 Peak Frequency (Hz) 0.692 i1 Peak Frequency (Hz) 0.671 i2 Center Frequency (Hz) 0.604 i2 Center Frequency (Hz) 0.615 t7 Center Frequency (Hz) 0.583 t7 Center Frequency (Hz) 0.596 t7 Peak Frequency (Hz) 0.523 t7 Peak Frequency (Hz) 0.564

Variable PC1 Weight (g) 0.741 Culmen (mm) 0.694 Tarsus (mm) 0.634 Wing (mm) 0.298

179

Table 4-5. Correlations of univariate song frequency measures with a body size index (morphology PC1) for males from all sites. See chapter 3 for a detailed description and visualization of the measured frequency variables. Significant correlations are bolded and effect sizes are listed for significant correlations only. Data are presented primarily for descriptive purposes to evaluate the relative effect of body size on the frequencies of different song components, following a significant multivariate effect; consequently we did not apply P-value corrections for multiple comparisons.

Pearson Correlation P effect size Selection and variable coefficient r value (r2 in %) i1 Peak Frequency (Hz) -0.01 0.475 - i1 Center Frequency (Hz) -0.13 0.284 - i2 Peak Frequency (Hz) -0.42 0.026 17.6 i2 Center Frequency (Hz) -0.30 0.088 - i3 Peak Frequency (Hz) -0.43 0.024 18.1 i3 Center Frequency (Hz) -0.39 0.038 14.8 i4 Peak Frequency (Hz) -0.47 0.013 22.3 i4 Center Frequency (Hz) -0.46 0.016 21.2 t2 Peak Frequency (Hz) -0.40 0.034 15.7 t2 Center Frequency (Hz) -0.44 0.019 19.7 t3 Peak Frequency (Hz) -0.39 0.039 14.8 t3 Center Frequency (Hz) -0.34 0.060 - t7 Peak Frequency (Hz) 0.14 0.273 - t7 Center Frequency (Hz) 0.16 0.244 -

180

Chapter 5. Does song function in territorial defense in the Western Sandpiper? An experimental approach.

5.1. Abstract

The function of bird song has been studied extensively in songbirds. Much less is known about the songs found in many shorebird species and what functional roles they fulfill. Whether these songs play a role in territorial defence has been a matter of dispute for some species, including the Western Sandpiper (Calidris mauri). A previous study suggested that male songs in this species function primarily in mate attraction, finding little support for a territorial role. In this study we combine behavioural observations with an experimental approach to revisit the role of song in territorial defence in Western Sandpipers. Our study also explores the general utility of song playback experiments in a monogamous shorebird. Western Sandpipers do not appear to re-nest within the same breeding season unless their clutch fails, extrapair paternity rates in this species are low and females appear to resist extra-pair copulation attempts. Thus, we did not expect males to continue song activity past the onset of incubation, if song functions solely in mate attraction, but did expect to see such activity if it plays a role in defending a territory or resource. Under the latter hypothesis, we also expected males to sing in response to song playbacks and to remain responsive to such playbacks throughout the breeding season. Behavioural observations revealed that both male and female Western Sandpipers sing during incubation and brood-rearing. Furthermore, song playbacks elicited song responses from both pre-incubation and incubating males. We therefore conclude that Western Sandpiper songs likely serve a dual function that includes a role in

181

territorial defence. Finally, our study also shows that Western Sandpipers respond to song playback and that their responses can be quantified.

5.2. Introduction

Bird songs frequently are solely or partially defined by their role in mate attraction and/or territorial defense (Spector 1994), but may serve additional functions such as mate guarding or communication with mates or offspring in a variety of contexts (Ritchison 1983; Langmore 1998; Marshall-Ball et al. 2006). Shorebirds can be highly vocal and the males of many species give complex vocalizations which have been termed songs and are frequently associated with display flights during the breeding season (Dixon 1918; Oring 1968; Bengtson 1970; Jehl 1973; Dabelsteen 1978; Nethersole- Thompson and Nethersole-Thompson 1979; Miller 1983; Miller et al. 1983; Miller et al. 1988; Whitfield and Brade 1991; Miller 1996). Hypotheses on the functions of these shorebird songs have been primarily based on limited behavioural observations and the phenology of their occurrence (e.g. Witherby 1948; Brown 1962; Oring 1968; Forsythe 1970; Parmelee 1970; Holmes 1973; Dabelsteen 1978; Nethersole-Thompson and Nethersole-Thompson 1979; Whitfield and Brade 1991). Their role in territorial defense in particular has been a matter of dispute (Parmelee 1970; Holmes 1973; Cramp 1983).

In Western Sandpipers (Calidris mauri), early observations of simultaneously displaying males and of songs given during agonistic interactions suggested a possible territorial function (Brown 1962) and Holmes’ (1973) behavioural observations led him to support a dual function. He also argued that the reduction in song rate after the pair formation period does not necessarily negate the territorial function hypothesis as it might be explained by well-established territories and time trade-offs with incubation and off- territory feeding (Holmes 1973). A more recent study also found that song rates were highest during pair formation in Western Sandpipers and that females did not perform song displays (Lanctot et al. 2000). Furthermore, that study established that a male’s

182

familiarity with his mate affected song rates – lending support to the mate attraction hypothesis, whereas familiarity with the site did not appear to do so – and hence did not support a territory defense function (Lanctot et al. 2000). These findings – along with observations from other studies of monogamous shorebird species (e.g. a lack of song displays during incubation and brood-rearing and by females, overlap of display areas) – led the authors to conclude that Western Sandpiper song displays likely function primarily in mate attraction (Lanctot et al. 2000).

In our study we combine behavioural observations with an experimental approach to revisit the role of song in territorial defense in Western Sandpipers. Certain aspects of the breeding biology of this species make it particularly well suited for this purpose. Western Sandpipers do not appear to re-nest within the same breeding season unless their clutch fails, extrapair paternity rates are low and females appear to resist extra-pair copulation attempts (Holmes 1972; Lanctot et al. 2000; Blomqvist et al. 2002a; Blomqvist et al. 2002b; Ruthrauff et al. 2009; Franks et al. 2014). Thus, males should restrict their attempts to attract females to the time period preceding the completion of a clutch or following the failure of such. If song functions solely in mate attraction, males should therefore have little cause to sing while attending an intact nest or brood. Brooding males in particular would be extremely unlikely to encounter receptive females, especially at the more northerly sites where the breeding season is shorter and re-nesting is rare (Sandercock et al. 1999; Ruthrauff et al. 2009). If song plays a role in defending a territory or resource, however, we would expect to see continued song activity during incubation and brood-rearing periods. The occurrence of song in both sexes during these periods would lend support to the territory defense hypothesis. Under this hypothesis, we would also expect males to sing in response to song playbacks and to remain responsive to such playbacks throughout the breeding season.

In addition to providing a means for elucidating the function of Western Sandpiper song, testing male responses to song playbacks also serves the purpose of

183

exploring the general utility and validity of song playback experiments in this species, with implications for similar studies in other monogamous shorebird species. If Western Sandpiper song has no territorial purpose, males might not respond to song playbacks in a way similar to that of most passerines. In establishing whether or not male Western Sandpipers respond to such playbacks and if their response can be measured quantitatively, our study also aims to lay the basic foundations for future song playback experiments.

5.3. Methods

5.3.1. Behavioural observations:

We observed Western Sandpipers at breeding sites near Wales (65°36'N, 168°05'W; May 29 – June 04, 2009), Kotzebue (66°50'N, 162°34'W June 11-24, 2009) and Nome (64°26'N, 164°56'W; May 21-26, June 07, 2009 and May 23 – July 1, 2010), Alaska. Individuals were observed and/or recorded during opportunistic encounters or checks on known nests or broods. We observed incubating males at each of the three sites, but only observed females and brooding individuals at the Kotzebue (one female) and Nome sites. We encountered males on a total of 52 nests in Nome (N=37), Wales (N=6) and Kotzebue (N=9). In some instances, song playback was used to elicit song responses, independent of the controlled playback experiments described below.

5.3.2. Playback experiments:

Subjects

We conducted playback experiments at the Nome breeding site in 2010 using male subjects, identified based on culmen length, genetic sex, sex of their mate (if known

184

female) or their behaviour (e.g. courting a female). Males with no obvious nest association encountered early in the season (May 20-29) were classified as pre-incubation males (N=12). Six of the pre-incubation subjects were observed in close proximity to a female, i.e. they were either paired or courting a female (e.g. displaying tail-up courtship stances (Holmes 1973)), while in the other six cases no female could be detected in the immediate vicinity. For playbacks with pre-incubation males we kept a minimum distance of 60m between subjects that were unmarked and could not be identified as a different individual. Males encountered on nests later in the season (June 10-18) were classified as incubating males (N=12).

Stimuli

We created playbacks from male songs recorded at the Nome site in 2009 with a Marantz PMD 670 recorder and a Sennheiser ME62 microphone. We selected high quality recordings with good signal-to-noise ratios. Black-throated Green Warbler songs served as controls (heterospecific song). As none of the songs contained frequencies below 1kHz, we filtered the recordings to remove low-frequency noise below this threshold. We created the playback stimuli in Adobe Audition 3.0 (Adobe Systems, San Jose, California) and each playback stimulus consisted of three songs of either type spaced and flanked by 3s intervals of silence.

Playback procedure and data collected

Each male subject was exposed to both types of stimuli: Western Sandpiper songs and controls and different songs and controls were used for each individual within breeding and non-breeding groups. We pseudo-randomized the order in which the stimulus types were presented, i.e. we randomized the order but with the restriction that both types had to be presented first equally often within the two subject groups (pre- incubation and incubating males). A portable loudspeaker and a three-dimensional wooden Western Sandpiper decoy (Fig. 5-1; present for both intra- and heterospecific (control) songs) were placed at an average distance of 7m from a subject, with the decoy

185

initially obscured by a black shroud. Playback experiments consisted of a 30s pre- playback observation period, uncovering of the decoy, playback of the first stimulus type with a 60s playback observation period, a pause (minimum 6min) during which the decoy was covered and moved to a different position, followed by another 30s pre-playback observation period and the playback of the second stimulus type with its 60s observation period. If males were observed to directly interact with another male during the initial pre-playback period, we delayed the playback and started a new pre-playback period to avoid possible confounding effects of the interaction.

Data collection and statistical analysis:

We used a Marantz PMD 670 recorder and a parabola-mounted Sennheiser ME62 microphone to record and quantify song responses. These recordings also allowed us to measure the response latency from spectrograms in Raven 1.4 (Hann window size 312 points, 90% overlap, grid spacing 46.9Hz). We defined response latency as the time between the onset of the first song in the playback to the onset of the first song response. If no song response occurred, response latency was set to 60s (the duration of the observation period). We noted whether males responded before the end of the playback song thereby overlapping their song with the playback song. Additionally, we recorded whether males directly approached the decoy, displayed agonistic signals or initiated agonistic interactions with the decoy or other birds in the immediate vicinity. Following Holmes (1973), we considered wing-up displays, upright postures and crouched posture approaches or approaches with chase note vocalizations to be agonistic signals. If a male made physical contact with or actively chased other birds we classified this as an agonistic interaction. We calculated a song number differential by subtracting two times the number of songs sung during the pre-playback observation period (30s) from the number of songs sung during the playback observation period (60s). We used a Wilcoxon signed rank test to compare song number differentials and response latencies in the two treatments (Western Sandpiper playback versus control playback) for pre-incubation and

186

incubating males. We also compared the responses of the two male groups to Western Sandpiper song with a Kolmogorov-Smirnov test.

5.4. Results

5.4.1. Behavioural observations:

Both male and female Western Sandpipers sang during incubation and brood- rearing. Males sang on the ground, in display flight and during chases. We observed singing behaviour in 87% of the incubating males that we encountered on or near their respective nests. Many of these males sang on multiple occasions. In addition, we observed song activity from a minimum of 12 brood-rearing males. The presence of unmarked males along with the high mobility of family units made it difficult to distinguish among individual brooding males. We therefore report only the minimum absolute number of singing males at the brooding stage.

5.4.2. Playback experiments:

Playbacks of Western Sandpiper song elicited a song response from 23 out of 24 males. Males encountered prior to incubation onset responded with up to 5 songs (Mdn=2.00, IQR=1.25-3.75; Fig. 5-2A) to Western Sandpiper song playback, which was significantly higher than to control playbacks (Mdn=0.00, IQR=0.00-0.00; T=0, P<0.001, r=-0.6028). Incubating males also responded to Western Sandpiper songs, but with a maximum of 2 songs (Mdn=1.00, IQR=1.00-1.75; Fig. 5-2A), and did not respond to control playbacks (Mdn=0.00, IQR=0.00-0.00; T=0, P<0.001, r=-0.64598). While the median number of songs was lower in incubating males, this difference was not significant (Kolmogorov- Smirnov test, Z=1.01, P=0.115).

187

Males responded quickly, with a median response latency of 6.21s for pre- incubation males (IQR=2.10-11.15s, control Mdn =60.00s, IQR=60.00-60.00s; T=1, P<0.001, r=-0.58076) and 2.32s for incubating males (IQR=1.77-8.08, control Mdn =60.00, IQR=60.00-60.00; T=0, P<0.001; r=-0.6245; Fig. 5-2B). The response latency in incubating males was not significantly lower than in pre-incubation males (Z=0.577, P=0.757). In 66.7% of trials, the responses overlapped with the end of at least one playback song, with some males overlapping more than one song (Table 5-1). The song responses of pre-incubation males frequently were accompanied by direct approaches and/or agonistic signals or interactions (Table 5-1). Incubating males, however, rarely displayed these behaviours (Table 5-1).

5.5. Discussion

While Lanctot et al. (2000) showed that Western Sandpiper song likely functions in mate attraction, our study suggests that it also functions in territorial or resource defense. Under the territorial defense hypothesis we predicted that males would sing in response to song playback and would continue to sing and respond to playback during incubation. These predictions were met. While a song response of pre-incubation males might occur under the mate attraction hypothesis if songs are perceived as an indicator of female presence, incubating males would be unlikely to respond under this scenario. In fact, such responses would likely be maladaptive as they could attract the attention of predators, potentially endangering to the nest. The occurrence of song in incubating and brooding females lends further support to the territory defense hypothesis. Male Western Sandpiper song therefore appears to – at the very least – have a dual function, but may have additional functions, as has been suggested for other shorebird species (e.g. Witherby 1948; Oring 1968; Nethersole-Thompson and Nethersole-Thompson 1979).

In Western Sandpipers as in many other shorebird species song rates are highest early in the breeding season (Dabelsteen 1978; Lanctot et al. 2000; Nebel and McCaffery

188

2003). While incubating males clearly responded to song playbacks, they did so with a lower median and maximum number of songs than pre-incubation males though this difference was not statistically significant. Well-established territory boundaries might facilitate lower song rates later in the breeding season (Holmes 1973). Additionally – and perhaps more importantly – incubating and brooding birds likely face trade-offs with parental care that would also lower song rates: parental care is, after all, time-consuming and requires cryptic behaviour to avoid attracting the attention of predators (Holmes 1973; Lattin and Ritchison 2009). Despite these trade-offs, the response of incubating males was equally as fast as that of pre-incubation males. The immediacy of the response of both pre-incubation and incubating males is also evident in the large percentage of overlapping songs. In some bird species song overlapping signals increased aggression or willingness to escalate; in others it may signal withdrawal from conflict (Vehrencamp et al. 2007; Naguib and Mennill 2010). Interactive playback experiments might help to determine its communicative significance in Western Sandpipers. As our study shows that Western Sandpipers respond to song playback and their responses can be quantified, further playback studies can be conducted. Acoustic communication in shorebirds remains poorly studied and playback experiments are few, but show great potential: they have been successfully used to investigate chick responses to different types of parental vocalizations, to study responses to distress and alarm calls, to test the communicative significance of specific song features and to show discrimination among songs of different subspecies (Heidemann and Oring 1976; Leger and Nelson 1982; Douglas 1998; Haase 2002; Johnson et al. 2008). This potential should be more fully explored to investigate various aspects of communicative behaviour in shorebirds.

5.6. References

Bengtson, S.-A. 1970. Breeding Behaviour of the Purple Sandpiper Calidris maritima in West Spitsbergen. Ornis Scandinavica 1:17-25.

189

Blomqvist, D., M. Andersson, C. Kupper, I. C. Cuthill, J. Kis, R. B. Lanctot, B. K. Sandercock, T. Szekely, J. Wallander, and B. Kempenaers. 2002a. Genetic similarity between mates and extra-pair parentage in three species of shorebirds. Nature 419:613- 615.

Blomqvist, D., B. Kempenaers, R. B. Lanctot, and B. K. Sandercock. 2002b. Genetic parentage and mate guarding in the Arctic-breeding Western Sandpiper. Auk 119:228-233.

Brown, R. G. B. 1962. The Aggressive and Distraction Behaviour of the Western Sandpiper Ereunetes mauri. Ibis 104:1-12.

Cramp, S., Ed. 1983. Handbook of the Birds of Europe, the Middle East and North Africa: The Birds of the Western Palearctic. Oxford University Press, Oxford.

Dabelsteen, T. 1978. An Analysis of the Song-Flight of the Lapwing (Vanellus vanellus L.) with Respect to Causation, Evolution and Adaptations to Signal Function. Behaviour 66:136-178.

Dixon, J. 1918. The Nesting Grounds and Nesting Habits of the Spoon-Billed Sandpiper. Auk 35:387-404.

Douglas, H. D. I. 1998. Response of Eastern Willets (Catoptrophorus s. semipalmatus) to vocalizations of Eastern and Western (C. s. inornatus) willets. Auk 115:514-518.

Forsythe, D. M. 1970. Vocalizations of the Long-Billed Curlew. Condor 72:213- 224.

190

Haase, B. 2002. The Use of Tape-Recorded Distress Calls to Increase Shorebird Capture Rates. Wader Study Group Bulletin 99:58-59.

Heidemann, M. K., and L. W. Oring. 1976. Functional Analysis of Spotted Sandpiper (Actitis macularia) Song. Behaviour 56:181-193.

Holmes, R. T. 1972. Ecological Factors Influencing the Breeding Season Schedule of Western Sandpipers (Calidris mauri) in Subarctic Alaska. American Midland Naturalist 87:472-491.

Holmes, R. T. 1973. Social Behavior of Breeding Western Sandpipers Calidris mauri. Ibis 115:107-123.

Jehl, J. R. 1973. Breeding Biology and Systematic Relationships of Stilt Sandpiper. Wilson Bulletin 85:115-147.

Johnson, M., S. Aref, and J. R. Walters. 2008. Parent-offspring communication in the Western Sandpiper. Behavioral Ecology 19:489-501.

Langmore, N. E. 1998. Functions of duet and solo songs of female birds. Trends in Ecology & Evolution 13:136-140.

191

Lattin, C., and G. Ritchison. 2009. Intra- and Intersexual Functions of Singing by Male Blue Grosbeaks: the Role of Within-Song Variation. Wilson Journal of Ornithology 121:714-721.

Leger, D. W., and J. L. Nelson. 1982. Effects of contextual information on behavior of Calidris sandpipers following alarm calls. Wilson Bulletin 94:322-328.

Marshall-Ball, L., N. Mann, and P. J. B. Slater. 2006. Multiple functions to duet singing: hidden conflicts and apparent cooperation. Animal Behaviour 71:823-831.

Miller, E. H. 1983. The Structure of Aerial Displays in 3 Species of Calidridinae (Scolopacidae). Auk 100:440-451.

Naguib, M., and D. J. Mennill. 2010. The signal value of birdsong: empirical evidence suggests song overlapping is a signal. Animal Behaviour 80: e11-e15.

Nebel, S., and B. J. McCaffery. 2003. Vocalization activity of breeding shorebirds: documentation of its seasonal decline and applications for breeding bird surveys. Canadian Journal of Zoology-Revue Canadienne De Zoologie 81:1702-1708.

192

Nethersole-Thompson, D., and M. Nethersole-Thompson. 1979. Greenshanks. Buteo Books, Vermillion.

Oring, L. W. 1968. Vocalizations of Green and Solitary Sandpipers. Wilson Bulletin 80:395-420.

Parmelee, D. F. 1970. Breeding Behavior of the Sanderling in the Canadian High Arctic. Living Bird 9:97-146.

Ritchison, G. 1983. The Function of Singing in Female Black-Headed Grosbeaks (Pheucticus melanocephalus) - Family-Group Maintenance. Auk 100:105-116.

Ruthrauff, D. R., J. N. Keller, and D. J. Rizzolo. 2009. Ecological factors regulating brood attendance patterns of the Western Sandpiper Calidris mauri. Ibis 151:523-534.

Sandercock, B. K., D. B. Lank, and F. Cooke. 1999. Seasonal declines in the fecundity of arctic-breeding sandpipers: different tactics in two species with an invariant clutch size. Journal of Avian Biology 30:460-468.

Spector, D. A. 1994. Definition in Biology: The Case of “Bird Song”. Journal of Theoretical Biology 168:373-381.

Vehrencamp, S. L., M. L. Hall, E. R. Bohman, C. D. Depeine, and A. H. Dalziell. 2007. Song matching, overlapping, and switching in the banded wren: the sender's perspective. Behavioral Ecology 18:849-859.

Whitfield, D. P., and J. J. Brade. 1991. The Breeding Behavior of the Knot Calidris canutus. Ibis 133:246-255.

193

Witherby, H. F., Ed. 1948. The handbook of British birds. H.F. & G. Witherby Ltd., London.

194

5.7. Tables

Table 5-1. Percentage of playback songs that were overlapped by male song responses. In 66.7% of the 24 trials males overlapped with one or more playback songs by starting to sing prior to the end of the playback songs.

Trials Overlap with playback songs No. % None 8 33.3% One 11 45.8% Two 4 16.7% Three 14.2%

Table 5-2. Occurrence of agonistic behaviour or direct approaches in response to song playbacks. Agonistic behaviour included wing-up displays, upright postures, crouched posture approaches, approaches with chase note vocalizations, physically contacting or chasing other birds. Pre-incubation males (N=12) frequently approached the decoy and displayed agonistic behaviour, incubating males (N=12) rarely did.

Agonistic Approach behaviour Male status No. % No. % pre-incubation 9 75.0 6 50.0 incubating 4 33.3 1 8.3

195

5.8. Figures

Figure 5-1. Photograph of the wooden decoy used in the playback experiments. The decoy was placed on top of the loudspeaker during the experiments.

196

197

A. B.

Figure 5-2. Response of male Western Sandpipers to song playback. Responses are shown for pre-incubation and incubating males. Each playback stimulus consisted of three songs and each male was subjected to both treatments: Western Sandpiper song and control playbacks (Black-throated Green Warbler song). Boxes depict the interquartile range (IQR) with the median line, whiskers the range, with outliers denoted as stars (>3xIQR). A.: Number of songs given by males in response to playback; the differential was obtained by subtracting two times the number of songs given during the 30s pre-playback period from the number of songs given during the 60s post-playback period. Males responded to Western Sandpiper song, but not to controls. The median number of songs in incubating males was lower than in pre- incubation males but not significantly so. B. Latency (s) between the onset of the first playback song and the first song response. If no response occurred, the value was set to 60s. Males typically responded very quickly to playbacks of Western Sandpiper song.

Chapter 6. Do male Western Sandpipers discriminate among local and non-local songs?

6.1. Abstract

Signals that are involved in mate attraction, recognition and/or choice possess the inherent potential to play an important role in restricting gene flow across populations if they differ among these populations. However, in order for such signal differences to be biologically significant (in the sense that they affect gene flow), they have to be of sufficient magnitude to affect mate choice. In this study, we investigate the biological meaningfulness of geographic variation in male songs of a shorebird, the Western Sandpiper, which shows evidence of divergence not only in song traits but also in genetic markers across its breeding range. We conducted playback experiments at a centrally located breeding site to evaluate whether the inter-site differences in song characteristics are sufficiently large to affect communication between individuals from different sites and thus have any potential to reduce gene flow among geographically separate breeding populations. As experiments involving females were not feasible, we studied whether males discriminate between songs from different breeding sites. Playback experiments revealed a difference in the intensity or immediacy of male song responses as measured by the Principal Component representing the number of songs, the percentage of songs that overlapped and the response latency. Specifically, the intensity of the song response was greater when local songs were played than when either of two types of non-local playback was used. This effect size was moderate, but suggests that males discriminated to some degree among local and non-local songs. However, males appeared to also respond to non-local songs and we observed no difference in their approach/agonistic response. We therefore consider it unlikely that females show a strong mating preference for local males based on song alone, but cannot rule out a higher level of discrimination among females that could lead to a reduction in gene flow among sites.

198

6.2. Introduction

Signals that are involved in mate attraction, recognition and/or choice possess the inherent potential to play an important role in the emergence or maintenance of reproductive isolation (West-Eberhard 1983; Slabbekoorn and Smith 2002; Catchpole and Slater 2008; Pillay and Rymer 2012). Such traits often show striking differences even among closely related species and may signal species identity thereby preventing hybridization (Becker 1982; Miller 1982; Price 1998; Ritchie et al. 1999). Consequently, intraspecific differences in these traits may be early indicators of past or ongoing divergence between populations or subspecies.

However, in order for such differences to be biologically significant (in the sense that they affect gene flow), they have to be of sufficient magnitude to affect mate choice (Ritchie et al. 1999). Even within a population or a certain locality, signals will vary to some degree, e.g. due to individual, motivational, seasonal or environmental differences, and individuals would be expected to show a certain tolerance level to accommodate intraspecific variation within certain limits (Dabelsteen and Pedersen 1985; Nelson and Marler 1990). Consequently, differences that are relatively small may not fall outside of the range of tolerance and hence may not affect mate choice (Pillay and Rymer 2012). Geographical differences do not always elicit a measurable difference in response in conspecifics (e.g. Nelson 1998), and even among species, responses to heterospecific signals sometimes occur and may lead to hybridization (e.g. Ryan and Rand 1995; Qvarnström et al. 2006; Wyman et al. 2011). Conversely, even seemingly small differences between populations or species can elicit a difference in response, making it difficult to evaluate the isolating potential of divergent mating signals based on the magnitude of their differences alone (Becker 1982; Seddon and Tobias 2010). Whether individuals in one geographical location discriminate among local and foreign signaling traits may therefore give us important insights into the potential effects of existing signal differences on gene flow among the respective populations.

In birds, song plays an important role in territory defense and mate attraction and often varies geographically within a species (Becker 1982; Nelson and Marler 1990; Slabbekoorn and Smith 2002; Catchpole and Slater 2008). For songbirds (oscines), this geographic variation may

199

be facilitated by song learning and is not necessarily associated with genetic divergence (Slabbekoorn and Smith 2002; Price 2007; Catchpole and Slater 2008). For taxa with primarily innate vocalizations, on the other hand, geographical differences in song parameters might be more tightly linked with genetic divergence (Miller 1996; Price 2007).

One species for which such a tighter link might be suspected is the Western Sandpiper. Like the males of many other bird species, male Western Sandpipers establish territories during the breeding season and employ songs to attract potential mates (Holmes 1973; Lanctot et al. 2000). While these songs are overall structurally similar across different breeding sites, geographical variation exists in the form of quantitative differences in fundamental frequency and song duration parameters (Chapter 3). Given that, unlike many songbird species, shorebirds do not appear to possess learned vocalizations (von Frisch 1956; von Frisch 1959; Miller et al. 1983; Miller 1986; Douglas 1996; Miller 1996), the aforementioned geographical differences in song might reflect genetic divergence among populations and could conceivably act as barriers to gene flow. A moderately strong, inverse relationship between individual body size and fundamental song frequency characteristics in male Western Sandpipers (Chapter 4) provides one possible mechanism through which song divergence could potentially be linked to genetic divergence (via changes in morphology).

Despite this apparent link between morphology and song and the postulated lack of song learning in shorebirds, the origins and, perhaps even more importantly, the consequences of the geographical variation in male Western Sandpiper songs remain unclear. We cannot exclude the possibility that a certain amount of phenotypic plasticity may be present when it comes to song production. Such plasticity might allow males to modify their songs to some degree e.g. in response to local environmental conditions (Gross et al. 2010; Medina and Francis 2012), regardless of a male’s own origin and without necessarily falling outside of the recognition template of conspecific receivers. Even if the differences in song characteristics across sites primarily reflect morphological differences, the respective song characteristics may still fall within an acceptable range of variation and thus might elicit the same level of response in the intended receiver, e.g. in a female choosing a mate.

200

We therefore conducted playback experiments to evaluate whether the observed inter-site differences in song characteristics are sufficiently large to affect communication between individuals from different sites and thus have any potential to reduce gene flow among geographically separate breeding populations. We chose to conduct our playback experiments on male subjects as males have been shown to readily respond to song playbacks and their responses are quantifiable (Chapter 5).

6.3. Methods

6.3.1. Subjects and study site

We conducted our playback experiment at a breeding site near Nome, AK, in 2010 prior to and during the incubation period (64°26'N, 164°56'W; 30 May – 22 June). Three of our 18 male subjects were paired or actively courting a female, one male was encountered feeding with no obvious nest association and the remaining males were in the incubation stage. For males without individual markings and nest associations we kept a minimum distance of 60m between playbacks. Some of the Nome playback songs were recorded from unmarked males in the previous year (see below), thus it was theoretically possible that a male would be treated to a playback containing its own songs. We attempted to reduce the likelihood of this occurrence by using playbacks outside of the area in which they were recorded, as adult males frequently return to the same territories in subsequent years (Holmes 1971).

6.3.2. Stimuli

All playback songs were recorded in 2009 with a Marantz PMD 670 recorder and a Sennheiser ME62 microphone from males at three arctic breeding sites: in the Yukon- Kuskokwim-Delta (YK-Delta, 61°21'N, 165°07'W), near Nome (64°26'N, 164°56'W) and near Kotzebue (66°50'N, 162°34'W), Alaska. The approximate aerial distances between the Nome site and the YK-Delta and Kotzebue sites amount to 343km and 288km respectively. We used Adobe Audition 3.0 (Adobe Systems, San Jose, California) to create playback stimuli from high quality recordings with good signal-to-noise ratios and applied filtering to remove all noise below the

201

frequency range of the songs (< 1kHz). We effectively created three types of stimuli: the first type contained male songs that were recorded locally, at the Nome site, whereas the other two types contained songs from different breeding sites (YK-Delta and Kotzebue respectively). Within each category each stimulus contained three songs, separated and flanked by 5s intervals of silence. To avoid pseudo-replication, different songs were used for each stimulus and each stimulus was only used once during the experiment.

6.3.3. Playback procedure

Prior to the start of a playback session, we placed a portable speaker and a shroud- covered, three-dimensional wooden decoy (see Chapter 5, Fig. 5-1) at an average distance of 6m from the male subject. During each playback session, we consecutively presented all three types of stimuli to the male – resulting in three separate trials per male. The order in which the different stimulus types were played was pseudo-randomized, with all six possible orders equally represented within the study design. Each of the three trials consisted of a 30s pre-playback observation period, followed by the uncovering of the decoy and the playback of one stimulus type with a 60s playback/post-playback observation period. To reduce the likelihood of habituation or carry-over effects (Hopp and Morton 1998), we separated trials by a break of at least 7min duration during which the decoy was covered and moved to a different position.

6.3.4. Data collection and statistical analysis:

A previously conducted playback experiment indicated that Western Sandpiper males respond to song playback with song, but showed no song response to a decoy presented with a control playback (median change in number of songs = 0). It also indicated that their responses may include approaches to the decoy as well as displays of agonistic behaviour (Chapter 5).

We therefore collected measures related to the song response as well as several response variables concerning approach and agonistic behaviour. We recorded song responses with a Marantz PMD 670 recorder and a parabola-mounted Sennheiser ME62 microphone for one minute following the onset of the playback and for each trial determined the number of songs,

202

the percentage of songs that overlapped with playback songs and the response latency (time between onset of first playback song and onset of first song response). The latter two variables were measured using spectrograms in Raven 1.4 (Hann window size 312 points, 90% overlap, grid spacing 46.9Hz). If no song response occurred, response latency was set to 60s. Behavioural responses were noted in the field, but were also video-taped and confirmed by video analysis where necessary. We recorded the percentage of time a male spent directly approaching the decoy and/or displaying agonistic signals (wing-up displays, upright postures and chase note vocalizations, following Holmes (1973)), time spent within 2m of the decoy and delta distance (the difference between the initial and the minimum distance to the decoy).

For the statistical analysis, we calculated a song number differential by subtracting the pre-playback song rate (per minute) from the number of songs sung during the one minute playback observation period, referred to as number of songs from here on. No other adjustments were necessary as no directed approach or other agonistic behaviour was observed during pre- playback. Hopp and Morton (1998) pointed out that pooling vocalization related data with approach/behavioural response variables may lead to the appearance of two clusters in Principal Component Analysis (PCA). Indeed, our song-related parameters showed weak or no correlations with the variables describing approach and agonistic behaviour, while correlations within these two variable groups were consistently higher. We therefore chose to analyze the two types of variables separately by conducting two Principal Component Analyses (PCAs) which allowed us to address this problem while still following the general multivariate procedure recommended by McGregor (1992, 2000): one on the three song-related measures and one on the three approach/agonistic behaviour related measures. Based on the component loadings (Appendix Table 6-1), PC1 appeared to represent strength of response in either case and was therefore retained; it explained 63.0% of the variation in song-related variables and 75.1% in approach/ agonistic behaviour variables. We used Friedman’s ANOVA to test for differences among the three playback treatments (Nome, Kotzebue and YK-Delta songs). Significant findings were followed-up with a Wilcoxon signed-rank test and significance levels for multiple comparisons were adjusted with the sequential Bonferroni method (i.e. tests were considered

203

significant at or below P=0.017). This procedure represents a non-parametric alternative to an ANOVA that is followed up with a post hoc test (Corder and Foreman 2009; Field 2009).

6.4. Results

All 18 male subjects sang, or showed agonistic/approach behaviour during Nome song playbacks. For Kotzebue playbacks, song or agonistic/approach behaviour during playbacks was recorded in 13 males (72%), for YK-Delta song playbacks in 11 males (61%). Overall, nearly all males (89%) showed this type of behaviour during at least one non-local song type.

The type of playback used affected the strength and/or immediacy of the song response, but did not appear to influence the intensity of the approach and agonistic behaviour. The song response PC1, which represented the number of songs, latency and the percentage of overlapping songs differed significantly among song treatments (n=18, χ2=6.21, df=2, P=0.044; Fig. 6-1a). Males responded more strongly to local than to each of the non-local songs, while their response did not differ between the two types of non-local song (Nome-Kotzebue: T=31, P=0.015, r=- 0.359; Nome-YK-Delta: T=32, P=0.017, r=-0.351; Kotzebue-YK-Delta: T=61, P=0.489, r=- 0.010). The behavioural PC1 which represented approach and agonistic behaviour variables did not differ among treatments (n=18, χ2=0.33, df=2, P=0.860; Fig. 7-1b). Summary statistics for all univariate measures are given in Appendix Table 6-2.

6.5. Discussion

Male Western Sandpipers sang and showed agonistic behaviour or approached the decoy/speaker during both local and non-local songs playbacks. As we have shown in Chapter 5, the presence of the decoy alone is not sufficient to elicit an increase in these behaviours; therefore males appeared to respond to all three types of song playbacks. We did, however, detect a difference in the intensity or immediacy of the song response as measured by the principal component representing the number of songs, the percentage of songs that overlapped and the response latency. Specifically, the intensity of the song response was greater when local

204

songs were played than when either type of non-local playback was used. This effect size was moderate, but suggests that males discriminated to some degree among local and non-local songs. Approach and agonistic behaviour, on the other hand, did not appear to be affected by playback type and may not be directly linked to the intensity or immediacy of the song response. Longer latency might reflect difficulty in interpreting the signal on the part of the receiver (Cure et al. 2012). An initial hesitation in responding – caused by a longer signal processing time – could also result in fewer songs overlapping as well as fewer songs overall. It is possible that once the decision to respond is made, the strength of the subsequent agonistic and approach behaviour is not primarily driven by signal type (local vs. non-local), which could explain why we observed a difference for the song but not the approach/agonistic behaviour response.

Overall, most male subjects at the Nome site sang, or showed agonistic/approach behaviour during at least one type of non-local song, occasionally even showing a strong apparent response. The differences in song parameters between Nome and the other two sites we tested therefore do not appear to be sufficiently large to constitute a substantial barrier in communication among Western Sandpiper males. This stands in marked contrast to a similar experiment in Willets (Tringa semipalmata), where a response was noted in only 22% of trials when individuals of the eastern subspecies (Tringa semipalmata semipalmata) were exposed to western subspecies song (Tringa semipalmata inornata), indicating a much stronger discrimination (Douglas 1998). A major difference to our study, however, is that both sexes were included in the experiment. In fact, the study indicated that for Western Willet song response levels among female Eastern Willets were particularly low, with none of the females showing a song response (Douglas 1998).

Ideally, we would have conducted our playback experiments with female subjects as their level of response would constitute a more direct measure of the potential effect divergent songs might have on mate choice and gene flow in Western Sandpipers. However, the involvement of female subjects was impractical for a number of reasons. Unpaired female Western Sandpipers on the breeding grounds tend to be relatively cryptic, rendering it difficult to detect such individuals in the field (Johnson et al. 2010). Such an experiment would also necessitate a large number of individually marked females to allow us to positively confirm the sex of an individual.

205

In the field, males tend to be more conspicuous and more easily identifiable as males, particularly early in the breeding season (based on their behaviour, e.g. displays). Furthermore, and perhaps most importantly, an adequate evaluation of female responses appeared difficult given the current lack of knowledge on what might constitute a positive female response, i.e. how a female might signal her choice, in this species.

In birds, male responses to song playback are frequently used in an attempt to assess the degree of reproductive isolation among populations, subspecies or species (Salomon 1989; Irwin et al. 2001; Grant and Grant 2002a, b; Colbeck et al. 2010; Danner et al. 2011). However, the level of male response does not necessarily accurately predict female mate choice (Searcy and Brenowitz 1988; Searcy et al. 2002; Nelson and Soha 2004; Seddon and Tobias 2010; Danner et al. 2011). The dual function of song (Chapter 5) creates inherent differences in the potential costs of false responses among the sexes (Searcy and Brenowitz 1988; Ratcliffe and Otter 1996). Male receivers respond in a territorial defense context. Their costs of falsely responding are assumed to be relatively low (unnecessary energy expenditure), whereas not responding to a relevant signal may carry the risk of losing a territory (Searcy and Brenowitz 1988). Males might therefore be expected to err on the side of responding to a signal (Searcy and Brenowitz 1988). Female receivers, on the other hand, usually respond to male songs in a mate attraction context. A false response in this context might lead to matings with heterospecific or non-local males that potentially carry high fitness costs in terms of reduced offspring fitness (Searcy and Andersson 1986; Searcy and Brenowitz 1988; Ratcliffe and Otter 1996). Consequently, females are often predicted to be more discriminative than males particularly when tested in a mate attraction or choice context (Searcy and Brenowitz 1988; Danner et al. 2011). In principle, this may more broadly apply to similar contexts across other taxa and – with some exceptions – seems to generally be supported by data (Sullivan and Leek 1987; Searcy and Brenowitz 1988; Bretagnolle and Robisson 1991; Searcy et al. 2002; Seddon and Tobias 2010; Danner et al. 2011; for exceptions see Nelson and Soha 2004) though in many cases the difference in experimental protocol for the two sexes may prevent a definitive conclusion (Searcy and Brenowitz 1988; Ratcliffe and Otter 1996; Riebel et al. 2002).

206

Even if females tend to be somewhat more conservative, playbacks involving male subjects can nevertheless provide us with a useful estimate of the isolating potential of divergent songs in many species (Grant and Grant 2002b; Patten et al. 2004; Balakrishnan and Sorenson 2006). Overall, males would be expected to show adaptive responses, i.e. signals that elicit a (strong) response should include those of potential rivals for female attention and exclude those that are unlikely to represent a threat, i.e. songs that females are unlikely to respond to (Salomon 1989; Melman and Searcy 1999; Grant and Grant 2002a, b; Balakrishnan and Sorenson 2006). Therefore, the degree to which males respond to non-local signals should at least provide an indication of whether or not females are likely to discriminate strongly against non-local songs (Balakrishnan and Sorenson 2006).

Nearly all males in our study sang, or showed agonistic/approach behaviour during playbacks of at least one non-local song type. This suggests that songs from breeding sites in the YK-Delta and near Kotzebue do not generally fall outside of the recognition template of males at the Nome breeding site. Given the strong tendency of males to show this apparent response to non-local songs along with the relatively minor differences in the strength of the response compared to local songs, we consider it unlikely that females show a strong mating preference for local males based on song alone. Our conclusion is further supported by the apparent low levels of genetic population structure across the breeding range (Chapter 2) and previously suggested low natal philopatry (Holmes 1971; Johnson and Walters 2008). Given, however, that males appeared to discriminate among local and non-local songs to some degree it is possible that a subtle bias in female mate choice may be present and could contribute to the low, but significant genetic population structure in this species.

6.6. References

Balakrishnan, C. N., and M. D. Sorenson. 2006. Song discrimination suggests premating isolation among sympatric indigobird species and host races. Behavioral Ecology 17:473-478.

207

Becker, P. H. 1982. The Coding of Species-Specific Characteristics in Bird Sounds. Pages 214-252 in Acoustic Communication in Birds, vol. 1: Production, Perception and Design Features of Sounds (D. E. Kroodsma, and E. H. Miller, Eds.). Academic Press, New York.

Bretagnolle, V., and P. Robisson. 1991. Species-specific recognition in birds - an experimental investigation of Wilson's storm-petrel (Procellariiformes, Hydrobatidae) by means of digitalized signals. Canadian Journal of Zoology-Revue Canadienne De Zoologie 69:1669- 1673.

Catchpole, C. K., and P. J. B. Slater. 2008. Bird song: biological themes and variations. Cambridge University Press, Cambridge, UK.

Colbeck, G. J., T. S. Sillett, and M. S. Webster. 2010. Asymmetric discrimination of geographical variation in song in a migratory passerine. Animal Behaviour 80:311-318.

Corder, G. W., and D. I. Foreman. 2009. Nonparametric Statistics for Non-Statisticians: A Step-by-Step Approach. John Wiley & Sons, Inc., Hoboken, N. J.

Cure, C., N. Mathevon, R. Mundry, and T. Aubin. 2012. Acoustic cues used for species recognition can differ between sexes and sibling species: evidence in shearwaters. Animal Behaviour 84:239-250.

Dabelsteen, T., and S. B. Pedersen. 1985. Correspondence between Messages in the Full Song of the Blackbird Turdus merula and Meanings to Territorial Males, as Inferred from Responses to Computerized Modifications of Natural Song. Zeitschrift für Tierpsychologie 69:149-165.

Danner, J. E., R. M. Danner, F. Bonier, P. R. Martin, T. W. Small, and I. T. Moore. 2011. Female, but Not Male, Tropical Sparrows Respond More Strongly to the Local Song Dialect: Implications for Population Divergence. American Naturalist 178:53-63.

208

Douglas, H. D. I. 1998. Response of Eastern Willets (Catoptrophorus s. semipalmatus) to vocalizations of Eastern and Western (C. s. inornatus) willets. Auk 115:514-518.

Field, A. P. 2009. Discovering statistics using SPSS: Introducing statistical methods., 3 ed. SAGE, London.

Grant, B. R., and P. R. Grant. 2002a. Lack of premating isolation at the base of a phylogenetic tree. American Naturalist 160:1-19.

Grant, B. R., and P. R. Grant. 2002b. Simulating secondary contact in allopatric speciation: an empirical test of premating isolation. Biological Journal of the Linnean Society 76:545-556.

Gross, K., G. Pasinelli, and H. P. Kunc. 2010. Behavioral Plasticity Allows Short-Term Adjustment to a Novel Environment. American Naturalist 176:456-464.

Holmes, R. T. 1971. Density, Habitat and the Mating System of the Western Sandpiper (Calidris mauri). Oecologia 7:191-208.

Holmes, R. T. 1973. Social Behavior of Breeding Western Sandpipers Calidris mauri. Ibis 115:107-123.

Hopp, S. L., and E. S. Morton. 1998. Sound Playback Studies. Pages 323-352 in Animal Acoustic Communication: Sound Analysis and Research Methods (S. L. Hopp, M. J. Owren, and C. S. Evans, Eds.). Springer-Verlag, Berlin/Heidelberg.

Irwin, D. E., S. Bensch, and T. D. Price. 2001. Speciation in a ring. Nature 409:333-337.

209

Johnson, M., D. R. Ruthrauff, B. J. McCaffery, S. M. Haig, and J. R. Walters. 2010. Apparent Survival of Breeding Western Sandpipers on the Yukon-Kuskokwim River Delta, Alaska. Wilson Journal of Ornithology 122:15-22.

Johnson, M., and J. R. Walters. 2008. Effects of mate and site fidelity on nest survival of Western Sandpipers (Calidris mauri). Auk 125:76-86.

McGregor, P. K. 1992. Quantifying Responses to Playback: One, Many, or Composite Multivariate Measures. in Playback and Studies of Animal Communication (P. K. McGregor, Ed.). Plenum Press, New York.

McGregor, P. K. 2000. Playback experiments: design and analysis. Acta Ethologica 3:3- 8.

Medina, I., and C. D. Francis. 2012. Environmental variability and acoustic signals: a multi-level approach in songbirds. Biology Letters 8:928-931.

Melman, D. S., and W. A. Searcy. 1999. Microgeographic song discrimination in a nonterritorial passerine, the Boat-tailed Grackle. Condor 101:845-848.

Miller, E. H. 1982. Character and Variance Shift in Acoustic Signals of Birds. Pages 253- 295 in Acoustic Communication in Birds, vol. 1: Production, Perception and Design Features of Sounds (D. Kroodsma, and E. H. Miller, Eds.). Academic Press, New York.

Miller, E. H. 1986. Components of Variation in Nuptial Calls of the Least Sandpiper (Calidris minutilla; Aves, Scolopacidae). Systematic Zoology 35:400-413.

210

Nelson, D. A. 1998. Geographic variation in song of Gambel's white-crowned sparrow. Behaviour 135:321-342.

Nelson, D. A., and J. A. Soha. 2004. Male and female white-crowned sparrows respond differently to geographic variation in song. Behaviour 141:53-69.

Patten, M. A., J. T. Rotenberry, and M. Zuk. 2004. Habitat selection, acoustic adaptation, and the evolution of reproductive isolation. Evolution 58:2144-2155.

Pillay, N., and T. L. Rymer. 2012. Behavioural divergence, interfertility and speciation: A review. Behavioural Processes 91:223-235.

Price, T. 1998. Sexual selection and natural selection in bird speciation. Philosophical Transactions of the Royal Society of London Series B-Biological Sciences 353:251-260.

Price, T. 2007. Speciation in Birds. Roberts and Company Publishers, Greenwood Village, Colorado.

Qvarnström, A., J. Haavie, S. A. Saether, D. Eriksson, and T. Pärt. 2006. Song similarity predicts hybridization in flycatchers. Journal of Evolutionary Biology 19:1202-1209.

211

Ratcliffe, L., and K. Otter. 1996. Sex differences in song recognition. Pages 339-355 in Ecology and evolution of acoustic communication in birds (D. E. Kroodsma, and E. H. Miller, Eds.). Comstock/Cornell University Press, Ithaca, New York.

Riebel, K., I. M. Smallegange, N. J. Terpstra, and J. J. Bolhuis. 2002. Sexual equality in zebra finch song preference: evidence for a dissociation between song recognition and production learning. Proceedings of the Royal Society B-Biological Sciences 269:729-733.

Ritchie, M. G., E. J. Halsey, and J. M. Gleason. 1999. Drosophila song as a species- specific mating signal and the behavioural importance of Kyriacou & Hall cycles in D. melanogaster song. Animal Behaviour 58:649-657.

Ryan, M. J., and A. S. Rand. 1995. Female Responses to Ancestral Advertisement Calls in Túngara Frogs. Science 269:390-392.

Salomon, M. 1989. Song as a possible reproductive isolating mechanism between two parapatric forms. The case of the chiffchaffs Phylloscopus c. collybita and P. c. brehmii in the western Pyrenees. Behaviour 111:270-290.

Searcy, W. A., and M. Andersson. 1986. Sexual Selection And The Evolution Of Song. Annual Review of Ecology and Systematics 17:507-533.

Searcy, W. A., and E. A. Brenowitz. 1988. Sexual differences in species recognition of avian song. Nature 332:152-154.

Searcy, W. A., S. Nowicki, M. Hughes, and S. Peters. 2002. Geographic song discrimination in relation to dispersal distances in song sparrows. American Naturalist 159:221- 230.

Seddon, N., and J. A. Tobias. 2010. Character displacement from the receiver's perspective: species and mate recognition despite convergent signals in suboscine birds. Proceedings of the Royal Society B-Biological Sciences 277:2475-2483.

212

Slabbekoorn, H., and T. B. Smith. 2002. Bird song, ecology and speciation. Philosophical Transactions of the Royal Society of London Series B-Biological Sciences 357:493-503.

Sullivan, B. K., and M. R. Leek. 1987. Acoustic Communication in Woodhouse's Toad (Bufo woodhousei). II. Response of Females to Variation in Spectral and Temporal Components of Advertisement Calls. Behaviour 103:16-26.

von Frisch, O. 1956. Zur Brutbiologie und Jugendentwicklung des Brachvogels (Numenius arquata L.). Z. Tierpsychol. 13:50-81.

von Frisch, O. 1959. Zur Jugendentwicklung, Brutbiologie und vergleichenden Ethologie der Limicolen. Z. Tierpsychol. 16:545-583.

West-Eberhard, M. J. 1983. Sexual selection, Social Competition, and Speciation. Quarterly Review of Biology 58:155-183.

Wyman, M. T., B. D. Charlton, Y. Locatelli, and D. Reby. 2011. Variability of Female Responses to Conspecific vs. Heterospecific Male Mating Calls in Polygynous Deer: An Open Door to Hybridization? PLoS ONE 6:6.

213

6.7. Figures 214

a. b.

Figure 6-1. Song (a.) and approach/agonistic behaviour (b.) responses to local and non-local song playbacks. PCAs were used to create a composite measure for the song response (a.) of male subjects and a separate measure for their approach and agonistic behaviour response (b.). In both cases, PC1 represented a measure of response strength and was therefore retained. The boxplots depict the responses to three different playback treatments: local songs recorded at the Nome site and non-local songs recorded at breeding sites near Kotzebue or in the YK-Delta, respectively (boxes depict the interquartile range (IQR) with the median line, whiskers the range, with outliers denoted as circles (>1.5xIQR) or stars (>3xIQR)). Each of the 18 male subjects was exposed to all three playback types. The strength of the song response differed among playback type with males responding more strongly to local than non-local song playbacks (Nome- Kotzebue: T=31, P=0.015, r=-0.508; Nome-YK-Delta: T=32, P=0.017, r=-0.497). Playback type did not, however, appear to affect approach and agonistic behaviour.

6.8. Appendix

Table 6-1. Component loadings for song response PC1 (a.) and approach and agonistic behaviour PC1 (b.). Based on these loadings both PC1s represented a measure of the strength of the respective response and were retained for further analysis. a. Song response PC1 Latency of response -0.824 Percentage of overlapping songs 0.791 Number of songs 0.765 b. Approach and agonistic behaviour PC1 Time spent within 2m of decoy 0.871 Delta Distance 0.866 Approach or agonistic behaviour (relative duration) 0.863

215

Table 6-2. Summary statistics for univariate response variables. Mean, Standard Deviation (SD), Median, 1st and 3rd Quartile (Q1, Q3) values for each of the measured response variables for the three playback types (n=18).

Time Relative Percent spent duration of Song of over- Delta within approach or playback Response Number lapping Distance 2m of agonistic type latency (s) of songs songs (m) decoy (s) behaviour Nome Mean 10.01 1.3 58.3 1.6 0.8 4.6 SD 18.52 0.8 42.9 2.8 2.4 6.9 Median 1.96 1.0 50.0 0.0 0.0 1.7 Q1 1.50 1.0 0.0 0.0 0.0 0.0 Q3 9.87 2.0 100.0 1.6 0.0 7.1 Kotzebue Mean 19.27 0.8 25.9 1.0 4.4 4.5 SD 26.24 1.0 36.7 2.2 13.0 9.0 Median 2.11 1.0 0.0 0.0 0.0 0.8 Q1 1.48 0.0 0.0 0.0 0.0 0.0 Q3 60.00 1.3 54.2 1.0 0.0 4.2 YK-Delta Mean 20.01 0.8 27.8 1.5 4.2 5.5 SD 25.92 1.6 42.8 2.1 9.8 13.9 Median 3.10 1.0 0.0 0.1 0.0 0.0 Q1 2.34 0.0 0.0 0.0 0.0 0.0 Q3 60.00 2.0 62.5 3.8 0.3 3.3

216