SPACE USE, HABITAT PREFERENCES, AND TIME-ACTMTY BUDGETS OF

NON-BREEDING DUNLIN (Calidris alpina pac@ca) IN THE FRASER RIVER

DELTA, B.C.

Philippa C.F. Shepherd

B.S., McGill University, 1989

M.S., Acadia University, 1994

THESIS SUBMITTED IN PARTIAL FULFILLMENT OF

THE REQUIREMENTS FOR THE DEGREE OF

DOCTOR OF PHILOSOPHY

In the Department

of

Biological Sciences

O Philippa C.F. Shepherd 2001

SIMON FRASER UNIVERSITY

July 200 1

Ail nghts reserved. This work may not be reproduced in whole or in part, by photocopy

or other rneans, without permission of the author.

uisiüons and Acquis i t i ie t raphic Setvices servicas bibliographiques

The author bas granted a non- exclusive licence allowing the National Lhrary of Canada to reproduce, loan, distriôute or sel1 copies of this thesis in microfoq paper or electronic formats.

The author retains ownership of the copyright in this thesis. Neither the thesis nor substantial extracts fiom it may be printed or otherwjse reproduced without the author's permission.

L'auteur a accordé une licence non exclusive permettant à la Bibliothèque nationale du Canada de reproduire, prêter, distribuer ou vendre des copies de cette thèse sous la forme & microficheifiùn, de reproduction sur papier ou sur format électronique.

L'auteur conserve la propriété du droit d'auteur qui protège cette thèse. Ni la thèse ni des extraits substantieis de celle-ci ne doivent être imprimés ou autrement reproduits sans son autorisation.

1 investigated aspects of the ecology of non-breedmg Dunlin (Calidris alpina pacifca) in

the Fraser River Delta, B. C., the most northerly site supporting a sizeable population in

winter (approximately 40,000 birds). 1 used radio telemetry, direct observation, and

Geographicai Information Systems to document site fidelity, space use patterns, habitat

preferences, and time-activity budgets of individual non-breeding Dunlin throughout the

24-hour day and twice-daily tidal cycles. Site fidelity and habitat preferences were

exarnined at both regional and local scales. Space use was quantified by estimating home

range and core area sizes, and by exarnining core area placement, macro-habitat choices

(marine versus terrestrial), and movement patterns within the home range. By following

individuals through time and calculating within-bird means by tide stage, macro-habitat,

and time of day, 1 minimized sampling biases and produccd a detailed picture of the

individuals' behaviour. Duntin were trapped in three areas within the Delta during two

non-breeding seasons (1995-96 and 1998), and categonzed by sex, and, where possible,

by age. 1 used a maximum likelihood mixture model to assign sex based on culmen

length. Dunlin were site faithful, both regionally (to the Fraser Delta) and locally (within

the Delta). 1 used compositional analysis to show that Dunlin chose habitats non-

randomly at both regional and local scales, and there were differences among sex and site

categones. Marine habitats were ranked highest. 1 assessed marine invertebrate prey

densities (large and small annelids, crustaceans, and molluscs) for intertidal micro-

habitats throughout the Delta, to examine their relationship with space use by Dunlin.

Across sites, marine home range size decreased as prey density within the home range

increased, with prey density accounting for 63% of the variance in home range size.

Within a single site, both marine home range and core area size decreased as prey density

increased, with prey density explaining 89% of the variance in home range size and 80%

of the variance in core area size. Dunlin marine core areas contained higher densities of

cnistaceans and small annelids than did the test ofthe home ranges. Most Dunlin also

used a range of terrestrial habitats, particularly at night. Soil-based agricultural crops

were preferred at a regional scale, and Pasture was the only agricultural crop that was

highly ranked and significantly preferred at both regional and local scales. Dunlin spent

on average at least 15.7 hours per 24-hour day foraging (depending on season), and at

least another 3 hours per day flying (measured in spring), leaving on average at most 5.3

hours per day for activities such as roosting, preening, vigilance, and aggression. The

percentage of tirne that Dunlin spent feeding did not differ between day and night, nor

between marine and terrestrial macro-habitats. Dunlin spent on average at least 7.1 hows

foraging at night, of which at least 2.9 occurred in terrestrial habitats, although the

relative use of marine and terrestrial habitats vacied considerably among individuals.

Fernales spent less tirne foraging than males, but there was no difference between age

classes. Finally, 1 compared the sex ratios and within-sex body sizes of DunIin wintering

in the Fraser River Delta with those wintering in central California, where ecological

variables might favour smaller birds. Both populations were similarly male-biased, even

though female Dunlin are larger than males, and 1 did not find significant within-sex size

differences between latitudes.

DEDICATION

To al1 the wild things that make my heart beat faster.

ACKNOWLEDGEMENTS

This thesis would not have been possible without the support of a great many people. First and foremost, 1 want to thank my prirnary thesis supervisor, Dov Lank. Dov's generosity is legendary among SFU students, and 1 know 1 speak for all of us in thanking him for his tirne, his ideas, his patient instruction, his thoughtful and careful editing, and for innumerable interesting discussions. His intellectual curiosity and enthusiasm for ideas are inspiring and infectious. 1 could always count on Dov to challenge me to go one step M e r than I would have on my own, and to think more deeply and more rigorousiy about everything under the sun. He has a way of asking questions that change the way 1 see things, and his guidance has fostered my growth as a scientist and as a person.

I thank Fred Cooke for acting as my senior supervisor a d for providing strong leadership to the Centre for Wildlife Ecology. His efforts have built the chair into a good home for dozens of graduate students, offerhg us the opportunity to work with and l e m from the research scientists and wildlife managers at the Canadian Wildlife Service as well as our staff and student culleagues within the SFU community. 1 thank the rernaining rnembers of my supervisory committee, Bob Elner, Rob Butler, and Ron Ydenberg, for their input and guidance over the years. 1 am grateful to Rob Butler for his careful reading of and comments on the tirst draff of the thesis. I am particularly gratefiil to Bob Elner for making a home for me at CWS, and for providing material, intellectual, and emotional support throughout my tenure. His wit, enthusiasm, and sense of humour make him a pleasure to work with.

I am greatly indebted to al1 the people who helped me with field work. I was blessed with a group of wondefil volunteers, several ofwhom came from the BCIT Fish and Wildlife Program. Unfortunately 1 did not keep a record of everyone who came out to help, but you know who you are and 1 thank you. In particular, 1 thank Kate Hagmeier, Oliver Busby, Julian Hudson, Terri Petersen, Graeme Fraser, Darren Lissimore, and Greg Mayne. They came out night after night, usually in the cold and rain, to help me try to catch Dunlin (oAen coming up empty-handed!). i was doubly blessed with a dedicated, hard-working, and good-humored field assistant, Lynn Campbell, for the 1997198 season.

1 am grateful to members of the agriçultural community in Delta, who kindly allowed me to capture and track shorebirds on their land. In particular, 1 thank the landowners on Westham Island who gave me access, at al1 hours of the day and night, to the private section of the dyke. i thank Roy Cuthbert for his friendship and for coming to my rescue more than once when my telemetry van died out on the dyke in the middle of the night. 1 thank John Ireland and Mary Taitt for their support and their permission to r o m freely in the Reifel Bird Sanctuary. Findly, I thank local naturalist Rick Swanson for diligently reporting al1 of bis interesting bird sightings in the Delta.

Without the help of many staff members at SFU and CWS, this thesis would have taken several more years to complete. I would like to thank Barry Smith, Evan Cwch,

Christine Hitchcock, and Brent Gurd for statistical advice and assistance. I thank Dwight McCullough, Kathleen Moore, Jason Kornaromi, and Steve Shisko, for al1 their technical assistance with Geographical Information Systems, and for providing me with the base maps of my study area. 1 thank Connie Smith and Barbara Sherman for their technical and administrative assistance. 1 thank Pm Krannitz for being a good Ladner fkiend and neighbour and for many restorative iunch-time walks at CWS. 1 thank Moira Lemon for answenng al1 my questions about the Fraser River Delta. 1 thank Pam Whitehead for her Corel Draw expertise and for executing the map in Figure 2.1. I thank Darlene Park, Shelagh Bucknell, Helene Marquis, and Helen Tjong at CWS, and Sylvia Foran, Brian Medford, Marlene Nguyen, Carol Conlin, Manoj Bakthan, and Tracy Lee ai SFU for their office and administrative support. 1 thank Doug Docherty for putting up with my corning and going from the CWS property at al1 hours of the day and night. 1 thank the lnterlibrary Loans staff for working so diligently to find the obscure articles and theses 1 requested.

1 was very lucky to be a part of the dynamic and dedicated graduate student community at SFU, especially the Wildlife Ecology and Behavioural Ecology Research Groups. In particular, I wish to thank Chris Guglielrno, Patrick O'Hara, Silke Nebel, Lesley Evans-Ogden, Heidi Regehr, Stephanie Hazlitt, Will Stein, Brett Sandercock, Doug Scharnel, Mark Drever, John Ryder, and Canna Gjerdrum for their friendship, support, and stimulating discussions over the years. I would also like to thank Nils and Sarah Wamock, and John, Susan, and Shane Kelly in California for their inspiration, advice, assistance, and intellectual contributions to my research. 1 thank East-Coasters Peter Hicklin, Sherman Boates, and Brian Hanington for fuelling my interest in shorebirds, and for their mentorship and support over the years.

1 have been blessed with a number of wonderful house-mates while living in B. C. First and foremost, 1 thank Andrea Hill, my Ladner roommate for five years, as weH as Andrea's Mum Patty, and her friend Laura, for welcoming me into their farnily and for taking such good care of me. 1 never would have suwived without you as my rock Andrea, and withoüt Shelby (Andrea's dog) to keep me Company al1 those long nights out in the field. 1 also thank the many long- and short-tenn members of the Odlurn household for making it such a fun place to live: Chris Guglielrno, Laura MacLean, Jane Astbury, Jason Nehring, Maki Sukita, Ian MacLean, Charles Dameau, Ji11 Cotter, and Lori Barjaktarovic. Tara Gill provided me with her fiiendship, as well as a physical and emotional refuge from the fray, and I've cherished my escapes to her Galiano Island home. Finally I would like to thank the awesome women in rny yoga class for keeping me sane and breathing, especially Charlene, the best teacher I've ever had.

1 have received rnuch love and moral support h m my thends back east (thank goodness for the flat rate long-distance phone plans!). 1 thank Eliza Griffiths, my soul- sister, for her loving voice on the phone, celebrating the good times with me and getting me through the bad ones. 1 thank Katherine Arkay for giving me the benefit of her expenence and her support when Z really needed it. 1 also thank the Sivananda Yoga Camp community in Val Morin, Quebec, for the restorative and enriching time 1 spent there on retreat. Traveling h m east to West, 1 wish to thank Teresa Swinamer, Madeleine Anderson, Lori Stewart, and Debbie van de Wetering. Your conversation and encouragement over the years kept me sane!

Most importantly, 1 thank my family for their unconditional and unwavenng love and support throughout my years of graduate studies. I thank my parents, Gyde and Rosemary, for their constant encouragement and for teaching me 1 c m do anything 1 set my heart and mind to. 1 thank my brothers, Gyde, Thomas, and Benedict, and my sister- in-law Mara for always being there when 1 need them and for reminding me there's life outside of school. 1 thank my niece Emily for providing her Dad with so many funny and heart-warming stories and for always putting a smile on my face. 1 thank my aunt Kathleen Barry for her love and kindness and great Company when 1 visit Montreal. 1 thank my cousin Benedict for welcoming me to the Benedictine monastery where he is a monk. Our nature-walks and all-encompassing conversations provided me with spiritual sustenance. 1 thank my godfather, Peter Fergusson, for his encouragement, advice, and generosity throughout my graduate work. 1 thank my godmother Anne Doran and her beloved late husband Jack Gibaut for always welcoming me with open arms and sharing my enthusiasm for the birds and the sea. Finally, 1 thank my late grandparents Hugh and Constance Findlay for their love and unwavenng pride in me, and 1 thank my late grandparents William and Frances Shepherd for the heritage (Cornu, our farnily cottage in the Laurentian Mountains) that fostered my love of nature and led me to this place.

In conclusion, 1 gratefully acknowledge the funding provided by Environment Canada (Fraser River Action Plan), the Natural Sciences and Engineering Research Council of Canada, the Research Network Program, the Canadian Wildlife Service, Simon Fraser University and the Centre for Wildlife Ecology, and the James L. Baillie Memorial Fund of the Long Point Observatory and Bird Studies Canada.

vii

TABLE OF CONTENTS

Title Page ............................................................................................................................. i . . Approval Page .................................................................................................................... il ... Abstract .............................................................................................................................. 111

Dedication ................................ ,.. ......................................................................................... v Acknowledgements .............................................................................................................. v ... ............................................................................................................. Table of Contents viii

List of Tables ....................................................................................................................... x List of Figures .................................................................................................................... xv . . List of Appendices .......................................................................................................... xvii

...................................................................................... Chapter 1: General Introduction 1 .......................................................................................................................... Introduction 2

Study Site ............................................................................................................................. 2 ...................................................................................................................... Study Species 3

Thesis Outline ..................................................................................................................... 4

Chapter II: Site fidelity and space use patterns of non-breeding Dunlin (Calidris alpina pacifica) in relation to prey density and potential predation risk in the Fraser River Delta. B.C. ................................................................ 8

.............................................................................................................................. Abstract 9 Introduction .......................................................................................................................... 9

.............................................................................................................................. Methods 14 ....................................................................................................... Statistical analyses 20

Results ................................................................................................................................ 27 .......................................................................................................................... Discussion 33

..................................................................................................................... Figure legend 44 ............................................................................................................................ Appendix 65

Chapter III: The importance of prey avltilability in determining the amount of space used by individual non-territorial Dunlin (Culidris

................................................................................................. alpina pacifica) in winter 68 .............................................................................................................................. Abstract 69

Introduction ........................................................................................................................ 69 Methods and statistical analyses ........................................................................................ 73

................................................................................................................................ Results 75 .......................................................................................................................... Discussion 77

............................................................................................................. Figure legend 8 I

Cbapter IV: Habitat preferences of Dualin (Calidris alpinapacifica) wintering in the Fraser River Delta, B . C ...................................................................... 90 Abstract .............................................................................................................................. 91

........................................................................................................................ Introduction 91 .............................................................................................................................. Methods 94

Statistical analyses ............................................................................................................. 94

Cbaptcr IV (con't) Results ................................................................................................................................ 98 Discussion ...................................................................................................................... 1 0 2 Figure legend ................................................................................................................... 1 08

Chapter V: Time-activity patterns and the importance of nocturnal foraging to higb-latitude temperaté wintering Dualin (Calidris alpina

......................................................................................................................... pac@ca) 1 15 .......................................................................................................................... Abstract 1 16

...................................................................................................................... Introduction 1 16 ............................................................................................................................ Methods 121

...................................................................................... Summary siatistics and analyses 124 ........................................................................................................... Results 128

................................................................................................................... Discussion 1 3 1

Chapter VI: Sex ratios of Dunlin wintering at two latitudes on tbe Pacific ............................................................................................................................ Coast

............................................................................................................................ Abstract ...................................................................................................................... Introduction

............................................................................................................................ Methods ............................................................................................................................. Results 163

........................................................................................................................ Discussion 163 ......................................................................................................... Acknow ledgements 1 6 6

............................................................................................................... Figure legend 1 6 7

Cbapter VII: Summary. management implications. and general . . . ................................................................................................................. implicatioas 1 71 Introduction ...................................................................................................................... 172 Site fidehty ....................................................................................................................... 173 Space use ...................................... .. ......................................... 173 Habitat Preferences .................................. .. ............................................................ 174 Time-activity budgets .................................................................................................... 177 Age effects ...................................................................................................................... 178

................................................................................................................... Sex effects 1 7 8 .......................................................................................................................... Sex ratios 180

..................................................................................................... Sub-population effects 180 ........................................................................................................ General implications 182

............................................................................................................... Literature cited 185

LIST OF TABLES

Table 2.1. Results of 2-way ANOVAs testing the effects of site, sex, and age categories on the size of overall, marine, and terrestrial Dunlin home ranges and core areas (HR = home range, CA = core area). Results of

.............................. reduced models reported where interactions not significant (P > 0.1) 45

Table 2.2. Least squares mean overall, marine, and terrestrial home range sizes 2 SE (km2) by site (taking sex into account), sex (taking site into account), and age (taking sex into account). Sites with shared letters not significantly different fiom each other (Bonferroni adjusted multiple t tests) .................. 46

TaMe 2.3. Least squares mean overall, marine, and terrestrial core area sizes + SE (km2) by site (taking sex into account), sex (taking site into account), and age (taking sex into account). Sites with shared letters not significantly different fiom each other (Bonferroni adjusted multiple t tests) .................. 47

Table 2.4. Least squares mean percentages of time (+ SE) Dunlin from each site (taking tide stage into account) spent in the terrestrial habitat at night (+ SE), by tide stage (taking sex into account). There was some indication that there may have been a difference arnong sites (FzJJ = 4.4, P = 0.06) ........................ 48

Table 2.5. Least-squares mean within-bird distance moved (km) between consecutive high and low tides (2 SE) by sex (taking site into account) and site (taking sex into account). For site comparisons, those with shared leners not significantly different fiom each other (Bonferroni adjusted multiple t tests) ................................................................................................................... 49

Table 2.6. Mean percent of 15-minute time blocks (+ SE) that Dunlin made small-scale (0.1-1 .O km) and large-scale (> 1 km) lateral movements (parallel to the shore, does not include movements in and out with the tides) by macro-habitat and time of day. N = number of Dunlin ..................................................... 50

Table 2.7. Least squares mean percent of 15-minute time blocks (2 SE) that Dunlin made small-scale (0.1 - 1 .O km) and large-scale (> 1 km) lateral movements by site (taking sex into account), sex (taking site into account), and age (taking sex into account). For site comparisons, those with shared tetters not significantly different fiom each other (Bonferroni adjusted

............................................................................ multiple t tests). N = number of Dunlin 5 1

Table 2.8. The results of tests to detemine whether there were differences among site, sex, or age categories of Dunlin in the densities (# per mZ) of marine invertebrates within their home ranges and core areas. The four invertehate types (LA = Large Annelids; SA = Small Annelids; C = C~s tacea ; M = Mallusca) were testai collectively using 2-way MANOVA (ALL), and separately using 2-way ANOVA. None of the nteraction temis wwe statistically significant at the P < 0.1 levet, so al1 results reported here are those of the reduced models ................................................... 52

Tabk 2.9. Least squares mean # invertebrat&m2 + SE (LA = Large Annelids; SA = Small Annelids; C = Crustacea; M = Mollusca) within Dunlin home ranges and cure areas, by site (taking sex into account), sex (taking site into account), and age [taking sex into account). For site comparisons, those with shared tenen not significantly different h m each other (Bonferroni adjusted multiple r tests) ...................................................... 53

Table 2.10. Paired t-tests comparing the mean densities of marine invertebrates within the home ranges versus the core areas of individual

................................................................................................................................ Dunlin 54

Table 2.1 1. Numbers of Dunlin and their predators, as well as Dunlin density estimates (numbers per km'of rnudflat), for the areas associated with each of the three banding sites in the Fraser River Delta. Count data are h m Christmas Bird Counts, 1995/96 through 1998199. Predators include falcons (PEFA, MERL, PRFA, GYFA, AMKE) and owls (SEOW, BAOW, SNOW) however, falcons are likely to be more important predators than owls (Chapter 5) .............................................................................................................. 55

Table 2.12. Comparisons of mean male and female body mass, body size, and percent time spent foraging between Dunlin from Wi and BB. included are resultç ofone-way ANOVAs comparing body masses and body sizes, as well as NO-way ANOVAs (taking tide stage into account) comparing foraging times, between sites within each sex. Foraging time means are least squares means taking tide stage into account ............................................................ 56

Table 3.1. Cornparison between territorial and non-territorial individuals in expected relative behavioural responses to areas differing in food density

................................................................................... and conspeci fic density 82

Table 3.2. Pearson correlation coefficients for invertebrates within Dunlin home ranges and core areas, al1 sites together and within the Mud Bay site alone ..................... 83

Table 3.3. Results of multiple regressions relating home range and core area size to the densities of four invertebrate types (LA = Large Annelids; SA = Small Annelids; C = Cnistacea; M = Mollusca), al1 sites together and within the Mud Bay site alone. PC1 = measure of body size (incorporating weight, wing length, and culmen length) ............................................................................................................. 84

Table 4.1. Ranking maûix for overall regional habitat preferences (al1 30 Dunlin). Values are the mean differences (2 SE) between the log usage ratios and the log availability ratios for each habitat in relation to every other habitat. Habitats were ranked in order of preference by counting the number of positive cases in each row of the matnx. Positive cases mean that the relative usage (use in relation to availability) of the habitat on the y-axis is greater than the relative usage of the habitats on the x-axis. The habitat with the most positive cases was ranked number 1 (most preferred) and the habitat with the least positive cases was ranked nurnber 8 (least preferred). Ranks that share a superscript are not statistically

................................. ...................................... different from each other (P > 0.05) .. 1 09

Table 4.2. Proportions of unconsolidated habitats available in the study area, mean proportions of unconsolidated habitats within Dunlin home ranges, and rank order of unconsolidated habitats within Dunlin home ranges (regional selection). Ranks that share a superscript are not statistically different from each other (P > 0.05). Ranks numbered 1 (most preferred) through 7 are selected, relative to ranks 8 through 12 which are avoided ............................................. 110

Table 4.3. Proportions of habitats available in the study area, mean proportions of habitats within Dunlin home ranges, and mean proportions of Dunlin radio locations found in each habitat. Overall, site- and sex- specific proportions reported, with sampie sizes in brackets .......................................... 1 1 1

Table 4.4. Rank order of habitats within Dunlin home ranges, and determined fiom Dunlin radio locations. Rank # 1 is the most preferred, and ranks that share a superscript are not statistically different from each other (P > 0.05). Overall and site-specific ranks are reported, with sample sizes in brackets .................... 1 12

Table 4.5. Results of MANOVA randomizations testing for regional and local habitat selection in Dunlin among sex and site categories. Regional selection was determined from the proportions of habitats within the home range, and availability from the proportions of habitats within the entire study area. Local selection was determined fiorn the mean proportion of radio locations within each habitat, and availability frorn the proportion of

............................................................................... each habitat within each home range 113

Table 5.1. Mean number of hours per day (+ SE) that individual radio-rnarked Dunlin spent feeding, by time of day and macro-habitat (M=marine, T=terrestrial) within each of the three sampling periods (winter 1995196, spring 1996, and spring 1998) ......................................................................................... 142

xii

Table 5.2. Least squares mean percent time Dunlin spent feeding (+ SE) by tide stage within each of the three sarnpling periods. N = number of . . . ....................................................................................................................... individuals. 143

Table 5.3. The effects on the percentage of time Dunlin spent feeding of time of day and macro-habitat (3-way ANOVAs with tide stage and sampling period), season and year (3-way ANOVAs with tide stage and sex), sex (3-way ANOVAs with tide stage and sampling period), and age (3-way ANOVA with tide stage and sex within the winter 1995196 sampling period) ............... 144

Table 5.4. Least squares mean percent time Dunlin spent feeding (+ SE) by time of day and by macro-habitat within each of the three sampling periods, taking tide stage into account. N = number of individuals ............................................ 145

Table 5.5. Results of four- and three-way repeated measures ANOVAs examining interactions between time of day and sex class, time of day and age class, and time of day and season in the percentage of time Dunlin spent

.............................................................................................................................. feeding 146

Table 5.6. Least squares mean percent time adult and juvenile D u n h spent feeding (f SE) in winter 1995196, by tide stage and taking sex into account. N = number of individuals .............................................................................................. 147

Table 5.7. Least squares mean percent time male and female Dunlin spent feeding (+ SE) by tide stage (H = high tide and L = low tide) within each of the three sarnpling periods. N = number of individuals .................................................. 148

Table 5.8. Mean percent time that radio-marked individual Dunlin (by telemetry) and unrnarked individual Dunlin (by focal sampling) spent feeding (+ SE), and mean percent of individuals in Dunlin flocks (by scan sampling) engaged in feeding (2 SE), overall and by tide stage in 1998 ........................ i49

Table 5.9. Mean percent time that radio-marked individuals (by telemetry, N = 15) and flocks (by observation) of Dunlin spent flying (2 SE), overall and by tide stage during the day in 1998 ......................................................................... 150

Table 5.10. Least squares mean percent time radio-marked male and female ............... Dunlin spent flying (+ SE) by tide stage in 1938. N = number of individuals 15 1

Table 5.11. Activity budgets of unrnarked individual Dunlin showing the mean percent time (5 SE) spent in each activity (focal sarnpling). N =

...................................................................................................... nurnber of individuals 1 52

Table 5.12. Activity budgets of Dunlin showing the mean percent of flock members (i SE) engaged in each activity (scan sampling), N = number of flocks ................................................................................................................................ 153

Table 5.13. Least squares mean body size variables, PCls, and overall % time feeding & SE) for male and female Dunlin used in the foraging time

................................. analyses, by sampling period and sex. N = number of individuals 154

Table 5.14. Number and percent of Dunlin confinned dead, presumed dead, and disappeared, by sex. N = number of radio-marked individuals

................................................................................................................. of known sex 1 55

xiv

LIST OF FIGURES

Figure 2.1. The Fraser River Delta study area ................................................................. -57

Figure 2.2. Study area showing banding sites (Wi = Westham Island, BI3 = Boundary Bay, and h43 = Mud Bay) and telemetry stations ................................... 58

Figure 2.3. Marine intertidal micro-habitats (as detennined by consolidating mapping data fiom Swinbanks 1977, Fisheries and Environment Canada 1977, Dunn et al. 1993, and McLaren and Ren 1995), and invertebrate collection transects on Roberts' Bank, in Boundary Bay, and in Mud Bay ....................................... 59

Figure 2.4. Dunlin radio locations detected during the day throughout the Fraser River Delta (N = 39 birds) ................................................................................ 60

Figure 2.5. Dunlin radio locations detected at night throughout the Fraser River Delta (N = 39 birds) ................................................................................... 6 1

Figure 2.6.95% home range (darkest polygon) and 30% core area (lightest polygon) of the Dunlin h m Westham Island carrying radio transmitter 6.059.

........................... The dots are the individual locations used to constnrct the home range 62

Figure 2.7.95% home range (darkest polygon) and 30% core area (lightest polygon) of the Dunlin from Boundary Bay canying radio transmitter 5.508+ The dots are the individual locations used to construct the home range ........................... 63

Figure 2.8.95% home range (darkest polygon) and 30% core area (lightest polygon) of the Dunlin from Mud Bay canying radio transmitter 5.884. The dots are the individual locations used to constmct the home range ........................... 64

Figure 3.1. Regressions showing the relationships between Durilin home ange s i x and the densities of each of the four invertebrate types (Large and small annelids, cnistaceans, and molluscs) acmss sites ..............................-............... 85

Figure 3.2. Regressions showing the relationships between Dunlin home ange size and the densities of each of the four invertebrate types (Large and mal1 annelids, crustaceans, and molluscs) within the Mud Bay site ......................... 86

Figure 3.3. Regressions showing the rehtionships between Dunlin core area size and the densities of each of the four invertebrate types (Large and small annelids, crustaceans, and molluscs) across sites ............................................. 87

Figure 3.4. Regressions showing the relationships between Dunlin core a r a size and the densities of each of the four invertebrate types (Large and smail annelids, crustaceans, and rnolluscs) within the Mud Bay site ......................... 88

Figure 3.5. Hypothesized trade-off between search ancilor travel tirne and conspecific interference resulting in a constant mean food intake rate across a

....................................................................................................... range of prey densities 89

Figure 4.1. Habitat base map showing categories used in the habitat ..................................................................................................... preference analyses 1 14

Figure 6.1. Culrnen distribution of Dunlin (males, fernales, and overall population) from the Fraser River Delta, British Columbia, estimated using mist-netted and reference samples (observed), and from Bolinas Lagoon, California estimated using the mist-netted sarnple (observed) and Bayesian prior probabilities ...................................................................................... 168

Figure 6.2. Monthly percent male Dunlin (including standard errors and sample sizes) for the Fraser River Delta, British Columbia and Bolinas Lagoon, California (Fraser River Delta sarnple corrected for sex differences in habitat use) ................................................................................................ 169

Figure 6.3. Hypothesized patterns of latitudinal clines in sex ratio (more males at higher latitude ends of each oval) in Dunlin wintering along the Pacific coast. Pattern 1 (solid line) may result if there is a latitudinal cline in sex ratio occurring in C. a. pacifica as a whole, and pattern 2 (dashed line) may result if there are latitudinal clines in sex ratio occurring in two partially geographically separate populations of C. a. pacificu ....................................... 170

xvi

LIST OF APPENDICES

Appendix 2.A.l. Mean # invertebrates/m2 i SE by transect and micro-habitat. N = number of cores. Transects and micro-habitats shown in Figure 2.3. The % total area values here are the proportion of the total area taken from the marine intertidal micro-habitat basemap (Figure 2.3) .................................................................. .65

Appendix 2.A.2. Mean # invertebratdcore + SE under dry and wet conditions in each habitat (LA = Large Annelids; SA = Small Annelids; C = Crustacea; M = Mollusca), and results of paired t-tests. N = number of cores .......................................... 66

Appendix 2.A.3. Mean # invertebratesicore + SE by time o f day in each habitat (LA = Large Annelids; SA = Small Annelids; C = Crustacea; M = Mollusca), and results of paired t-tests. N = number of cores ................................................................... 67

xvii

CHAMER ONE

GENERAL INTRODUCTION

INTRODUCTION

There has been considerable research into the winter ecology of shorebirds in

temperate coastal habitats over the past 30 years. In the last decade, and in the face of

declines in shorebird populations worldwide, data from these studies have been used to

develop predictive models of the potential consequences of habitat loss on shorebird

populations (Goss-Custard et ai. 1992,199Sa and b, 1998,2000, Yates et al. 1996,

Stillman et al. 1997 and 2000, Gill and Sutherland 2000, Stillman et al. 2000). For

example, data on the effects of interference competition on food intake rates at different

competitor densities have been used in behaviour-based models to predict how habitat

loss and the redistribution of birds into remaining habitats might affect swival (Stillman

et al. 1997 and 2000, Gill and Sutherland 2000, Goss-Custard et al. 2000).

Individuals, and classes of individuals, such as adults and juveniles, can Vary in

competitive abilities, so "an evolutionary understanding of the behaviour of individuals in

populations allows us to predict responses under changed conditions with greater

confidence than in the case of higher-level processes." (Sutherland & Gosling 2000, p. 4).

1 designed my study to answer basic questions about the winter ecology of Dunlin

(Calidris alpina pacifica) in the Fraser River Delta, British Columbia, that would be of

use for shorebird conservation and management. Specificaily, my objectives were: 1) to

use radio telemetry and GIS to quantifi site fidelity, space use patterns, habitat

preferences, and time-activity budgets of individual Dunlin of known sex, and, where

possible, age, fiom different sites within the Fraser River Delta, and 2) to integrate these

behavioural data with data on biotic factors such as the distribution of food and predatots,

and abiotic factors such as landscape features, tide stage, and time of day. By so doing, 1

can infer how the factors decting individual decisions create patterns at the population

level. Where appropriate, 1 use behavioural ecology theory to provide a framework and

to make predictions.

STUDY SITE

The Fraser River Delta in southwestern British Columbia is the largest wetland on

Canada's Pacific Coast and supports the country's highest densities of waterbirds, raptors

and shorebirds in winter (Butler & Campbell 1987). The Delta is also a key stopover site

for many species of migrant birds flying between breeding habitat in Canada, Alaska and

Russia and wintering habitat in southem USA and Central and South Arnerica. Over two

million shorebirds use the Fraser Delta annually, including internationally-important

populations of Dunlin (Calidris alpina) and Western Sandpipers (Calidris mazrri) (Butler

& Vermeer 1994, Butler, pers. comrn.). Westem Sandpipers and Dunlin stop over in the

Delta on both northward and southward migrations, and between 30 and 60 thousand

Dunlin are present for much of the non-breeding season (October-April) (Butler and

Vermeer 1994).

The Fraser Delta is approximately 10 kms south of the city of Vancouver, the

third-largest and one of the fastest-growing cities in Canada. Over the last 15 years there

has been considerable expansion of the human population in the ares with resultant

increases in housing, recreational, and industrial development. The river itself carries

agiculturai runoff and effluent from sewage treatment plants, paper mills, and other

industries out to sea via the Delta. An intensive Fraser River clean-up effort was initiateci

in 199 1 (Fraser River Action Plan), and, in 1994, a study was proposed to assess the

eficacy of Dunlin as a 'sentinel species' to gauge the health of the estuary ecosystem.

Shorebirds, and Dunlin in particular, had successfùlly been used as indicators of

environmental health in previous studies (Goss-Custard 1979; Goede and DeBniin 1985a

& 1985b; Burger 1986 & 1988; Goss-Custard and Moser 1988; Goss-Custard and Le V

dit Durrell IWO; Hill et al. 1993). One could measure, for example, levels of heavy

metals in Dunlin tissues, but for a meaningful interpretation of results, detailed

knowledge of their habitat use and site fidelity is necessary.

STUDY SPEClES

As many as nine subspecies of Dunlin have been identified worldwide, w o of

which, C. a. pacifka and C. a. hudsonia, occur in Cmda. Dunlin have a circumpolar

breeding range, and winter on or near coasts no& of the equator. Breeding habitat is

arctic and subarctic tundra, and they use coastal estuaries, intertidal flats, agicultural

lands, and interior seasonal wetlands during the non-breeding season (Butler and Vermeer

1994; Warnock and Gill 1996). C. a. pacifica breeds in Alaska and is a common winter

resident h m southem British Columbia to Mexico (Wamock and Gill 1996). The Fraser

River Delta is the northemmost site to support a large non-breeding population (> 10,000

birds) (Warnock and Gill 1996).

Despite the Dunlin's extensive geographical range, populations of several

subspecies appear to have declined in recent decades. Both the Canadian (2001) and U.

S. (2000) Shorebird Conservation Plans list Dunlin as a species of concern due to

declining populations (Donaldson et al. 2001, Brown et al. 2000). Research on European

wintering grounds has related Dunlin declines to man-made changes to wetland habitats

(Goss-Custard and Moser 1988). Wamock and Gill(1996) estirnate the loss of C. a.

pacifca winter habitat to be between 30 and Il%, and the U.S. Shorebird Conservation

Plan (Brown et al. 2000) attributes Dunlin declines to habitat loss dong the Pacific,

Atlantic, and Gulf coasts.

THESIS OUTLINE

This study investigates space use, habitat preferences, and time-activity budgets of

non-breeding individual Dunlin in the Fraser River Delta, B. C. Much of the researc h

into the non-breeding ecology of Dunlin and other shorebirds has taken place ai the

population level, though patterns aise primarily through individual behavioural

responses. With advances in telemetry technology and the miniaturization of radio

transmitters, researchers can now collect precise data through tirne on space use, habitat

preferences, and even tirne-activity budgets of small individuals over large areas. By

following individuals through time and caiculating within-bird means by tide stage,

macro-habitat, and time of day, 1 minimized sampling biases and produced a

comprehensive picture of the individuals' behaviour. With advances in Geographicd

Information Systems (GIS) software, it is possible to integrate the behavioural data with

data on the distribution of murces.

1 used individual Dunlin that différed by age, sex, and sub-population as study

subjects. "In order to predict dynamics of populations accurately, one must identifi

groups of individuals that differ in survival and reproductive values" (Lomnicki 1988, p.

86, quoted in Warnock 1994, p. 2). Differences in survivorship among groups cm aise

as a result of differences in morphology and behaviour. Birds of different sexes, ages,

and subpopulations can vary in such factors as body size, dominance status, experience,

and foraging strategy, variation that may result in differences in space use, habitat

selection, or activity among groups. Any such differences may or may not produce

performance inequalities, upon which natural selection can act. If performance does Vary

mong groups, we can make predictions about the potential effects of such differences on

survivorship and population dynamics. Warnock (1 994) determined that juvenile C. a.

pacifica had a tower probability of survival than adults, and postulated that ihis was due

to differences in experience between the age classes that resdted in predator-naïve

juveniles foraging in riskier habitat than adults. If performance does not Vary among

groups or individuals, such differences (in space use, habitat selection, activity) can

themselves reveal the range of behavioural strategies that exist in the population.

In chapter 2,1 examine the winter site fidelity and space use patterns of Dunlin at

three sites in the Delta, and investigate relationships with prey density and potential

predation risk. Space use was quantified by estimating home range and core area sizes,

and by exarnining core ma placement, macm-habitat choices (marine versus terrestrial

habitats), and movement patterns within the home range. 1 also assessed indices of the

relative quality of the three sites within the Delta using &ta on D d i n densities, prey

density, predator numbers, an index of disturbance, and indices of individual

performance. Chapter 2 contains a detailed description of the radio-telemeûy

methodology to which 1 refer in subsequent chapters.

in chapter 3,I examine the arnount of space used by individual, non-territorial

Dunlin in relation to prey density, prey patchiness, and body size. investigations of

factors affecthg space use by individual birds have largely been restricted to those

species exhibiting territoriality, due to the Iogistical difficulties of following individuals

that do not defend a perimeter or retain exclusive use of their spaces. 1 used the litetanire

on temtory size and its determinam to provide a theoretical fiamework, and

investigated: 1) whether individual non-territorial Dunlin use less space as marine

invertebrate density increases, and 2) whether estimated mean food intake rate by Dunlin

differs across a range of prey densitieslarea sizes. In other words, 1 examined whether

individuals employ space use strategies with regards to food (small ranges in high prey

density areas versus large ranges in low prey density areas), and, if so, whether the

strategy used affects the individual's performance, in terms of their rate of food

acquisition.

In chapter 4,1 quanti@ Dunlin habitat preferences at regional and local scales and

throughout the 24-hour day. Animals may use different criteria when making decisions at

different scales, so information h m more than one scale is needed to fidly understand

their requirements (Morris 1987, Pedlar et al. 1997, Saab 1999). Shorebirds, including

Dunlin, wintering in temperate estuanes are also active both day and night (Mouritsen

1994, Warnock and Takekawa 19%), and habitat use estimates based on data collected

only during the day may therefore be biased (Beyer and Haufler 1994).

In chapter 5,I describe my use of radio telemetry to determine how much time

individual Dunlin, wintering at the northern end of the non-breeding distribution in which

they are common, spend feeding each 24-hour day, and 1 evaluate the importance of

nochunal foraging. 1 collected data throughout the day and the nighî, in marine and

terrestrial macro-habitats, and through high and low tides, among which predation risk

and prey availability Vary. 1 also used daytime observational sampling to determine how

much time Dunlin spent on other activities, such as flying, roosting, preening, vigilance,

and aggression, by tide stage and macro-habitat.

In chapter 6,I compare populations of non-breeding Dunlin fiom two latitudes

dong the Pacific flyway, the Fraser River Delta, and Bolinas Lagoon, California to

determine whether, and to what degree, they exhibited sex ratios consistent with a

latitudinal cline. 1 also tested the hypothesis that the mean body size within each sex is

larger at the higher-latitude site (Fraser River Delta), where ecological variables might

favour larger birds. 1 describe a maximum Iikelihood mixture model that uses culmen

length distributions to estimate overall and monthly sex ratios for each population. 1 used

the same model to assign sex to the radio-marked Dunlin in the Fraser Delta.

Finally, in chapter 7,1 summarize my findings and discuss their implications for

the conservation and management of temperate wintering shorebirds, and Dunlin in the

Fraser River Delta in particular.

CHAPTER 2

SITE FIDELITY AND SPACE USE PATTERNS OF NON-BREEDING DUNLIN

(Calidris alpina pacifia) IN RELATION TO PREY DENSITY AND POTENTIAL

PREDATION RISK IN THE FRASER RIVER DELTA, B. C.

ABSTRACT

1 used radio telemetry to examine the winter site fidelity (within-year) and space

use patterns of Duniin ûapped at three sites in the Fraser River Delta, B.C., and

investigated relationships with invertebrate prey density and potential predation risk.

Space use was quantified by estimating home range and core area sizes, and by

examining cote area placement, macro-habitat choices (marine versus temûid), and

movement patterns within the home range. Birds from each of the three sites were radio-

marked, measured, categorized by sex, and, where possible, by age. Invertebrate prey

samples were collected dong transects at each site to quanti@ prey density. Finally, 1

assessed indices of relative site quality for Dunlin within the Delta using data on Dunlin

density, prey density, predator density, and indicators of individual performance. Dunlin

showed a high degree of within-year site fidelity, both regionally (to the Delta) and

locdly (to sites within the Delta). Dunlin home range and core area sizes differed among

the three sites, and were largest at the site where overall prey density was lowest. Dunlin

marine core areas contained higher densities of crustaceans and small annelids than did

their marine home ranges, indicating that Dunlin focussed their use of space on the better

feeding areas within their ranges. Femde Dunlin had larger marine home ranges than

males. Dunlin made littie use of the higher-risk terrestrial macro-habitat during the day

(1.6% of locations) when falcon species, their main predators, were active, but made

considerable use of that habitat at night (46.9 % of locations). Dunlin density was lowest

(102 birds/km2) at the site where prey density was lowest, and highest (192 birds/km2) at

the site where prey density was among the highest. Notwithstanding differences in

predator numbers, disturbance, and prey densities as well as Dunlin density and space use

among sites, 1 found no differences in Duniin performance indices (body condition and

percentage of t h e spent foraging).

INTRODUCTION

The use of space by migratory individuals cm be organized dong a continuum of

spatial and temporal scales (Myers 1984). Globally, over a year or more, individuals can

9

move thousands of kilometers between breeding and non-breeding sites. Regionally,

within the non-breeding season (not including migration), individuals can use space dong

a continuum of d e s fiom nomadism to site fidelity. Locally, within days to weeks,

individuals can switch habitats and can use more or less space and within their home

ranges. Patterns of food availability and predation risk can influence space use decisions

at al1 scales.

Regionally, fidelity to a wintering site can be advantageous if food is ~ ~ c i e n t l y

and dependably available. Familiarity with the distributions of predators rnay impmve

their ability to find food and avoid predation, and thereby decrease their likelihood of

mortality (Clark et al. 1993, Warnock 1994, Dierschke 1998). However, if food is

insuficient, or if there is variability in food availability, fidelity to that site would be less

advantageous and individuals may benefit fiom moving to altemate sites (Evans 1981,

Myers 1984, Cuadrado et al. 1995).

Familiarity with the distribution of predators and prey can also provide a benefit

locally, in terms of space use. Animals can use the knowledge gained fiom familiarity to

focus their use of space on areas of higher food availability and to make decisions about

if and when to switch habitats. They rnay specialize their foraging technique to improve

food intake (Quammen 1982). Individuals can also decrease their risk of predation

relative to that of the population as a whole by minimizing the use of more dangerous

parts of their space a d o r by modulating their behaviour while using riskier areas (Leger

and Nelson 1982, Kus 1985, Whitfield 1985, Warnock 1994).

Knowledge of site fidelity and space use patterns exhibited by a population rnay

help wildlife managers predict the effects of habiiat loss or alteration on that population.

Populations that rarely move rnay be more affected by the loss of a site, particularly if

alternative sites are rare, less productive, distant, andior already occupied. Populations

that naturally show more flexibility in site and space use rnay be l e s affected by natural

or anthropogenic changes to their habitat. Within a site, witdlife managers can assess the

relative importance of habitats by quantifjing space use patterns. Individuals rnay use

less space in areas where the habitat is more productive, and may also use some parts of

thei- space to greater or lesser extent than others.

1 used radio telemetry to quanti@ the within-year site fidelity and space use

patterns of individual Dunlin (Calidris alpina pacijca) wintering in the Fraser River

Delta, and Uivestigated relationships with prey density and ptential predation nsk. 1

quantified both regional and local site fidelity. To study space use patterns, 1 quantified

the amount of space used (home range and core area sizes), as well as core area location,

macro-habitat choices (marine versus terrestrial), and movements within the home range.

1 estimated these parameters using individual Dunlin of known sex, and, where possible,

age, trapped at three different sites within the Fraser Delta. Site fidelity and space use

patterns can Vary among individuais depending on their sex, age, or sub-population

affiliation, variation which has been atûibuted to differences in factors such as body size,

energetic requirements, dominance statu, experience, and foraging strategy (Schoener

1968, Smith and Evans 1973, Harestad and Bunnell 1979, Townshend 1981, van der

Have et al. 198 1, Hanington 1982, Ruiz et al. 1989, Wamock 1994, Warnock et al. 1995,

Dierschke 1998, Caldow et al. 1999, Mam 1999, McCloskey and Thompson 2000,

Relyea et al. 2000). If differences in site fidelity and space use patterns result in

performance differences among groups, predictions can be made about relative potential

non-breeding survivorship. If not, observations can be made about the range of

successhl behavioural strategies that have evolved in the population.

Site fidelity

Several species of shorebird, including Dunlin, can exhibit high levels of non-

breeding site fidelity (Pienkowski and Clark 1979, Evans 1981, Metcalfe and Furness

1985, Myers et al. 1988, Smith et al. 1992, Wamock et ai. 1995, Burton and Evans 1997,

Burton 2000). Clark et al. (1993) suggested that knowledge of the distribution of local

food resources contributed to the observation that resident Dunlin were less likely than

recent immigrants to die during cold spells. However, they did not exclude the

possibility that the recent immigrant Dunlin may have been in poorer condition than the

resident birds upon the immigrants' anivai at the new site.

Although Dunlin often exhibit high levels of fidelity to non-breeding sites

(Minton 1975, Pienkowski et al. 1979, Pienkowski and Clarke 1979, Ruiz et al. 1989,

11

Warnock 1994, Wamuck et al. 1995), they are also capable of behavioural flexibility

associated with variation in food availability (Ruiz et al. 1989, Warnock 1994, Warnock

et al. 1995). For example, M i n wintering in Boâega Bay, California were site faithful

from November through February 1987, but in March, approximately 75% of them made

crepuscular movemcnts ta destination(s) outside of the surrounding coastal and inland

sites (Ruiz et ai. 1989). lndividuals in the mobile group were in better body condition

han those that did not make crepuscular movements, despite higher travel costs, so they

may have had access to more profitable food resources elsewhere.

Marine invertebrates appear to be plentifid in the Fraser River Delta during winter

(McEwan and Farr 1986, Baldwin and Lovvom 1994). Wowever, there are occasional

pe rds of severe weather in which fomging habitat fieezes over (Baldwin and Lowom

1994). I examineci individual patterns of residency in the Delta to determine whether or

not Dunlin were faitfil to the estuary. 1 predict that Dunlin wintering in the Fraser Delta

should generally be site faithfid, asswning knowledge and experience provide a benefit in

terms of food acquisition, but that they should also be capable of distance movements

during periods of severe weather. In addition, I used data on the distribution of each

individual's radio locations throughout the Delta to examine local site fidelity within the

estuary. Again, assuming knowledge or expenence provide a foraging advantage, Dunlin

should be faithful to local sites when they are present in the Delta.

Space use patterns

An animal's home range is "the area in which it nonnally [ives, exclusive of

migrations, emigrations, dispersal movements, or unusual erratic wanderings" (sensu

Brown 1979, and core areas are "areas of concentrateci use within the home range"

(Samuel et al. 1985). Horne range size generaliy decreases with increasing habitat

productivity among vertebmte species (Schoener 1968, Harestad and Bunnell 1979, Peery

2000). Few studies have quantifid non-breeding home ranges in non-territorial

shorebird species (but see W m c k and Takekawa 1996, Plissner et al. 2000)' and none

have d i t l y related home range sizes to measurements of food resources. I measured

the density of intertidai invertebrates at three sites w i t h the Fraser Delta, and related

them to the home range and core area sizes of the Dunlin from each site. 1 predict that

Dunlin home ranges and core areas will be largest at the site where invertebrate density is

lowest, assuming p a t e r distance traveled between food patches where food density was

lower.

Dunlin and other shorebird species can also Vary the way they use the space

within their home range in response to variation in prey density and predation risk. Many

shorebird species regularly switch feeding habitats within their winter ranges in response

to changes in foraging profitability (Goss-Custard 1969, Kelly and Cogswell 1979, Page

et al. 1979, Connots et al. 198 1, Townshend 198 1, Colwell 1993, Rottenborn 1996,

Colwell and Dodd 1997). Some species, including Dunlin, move from intertidal areas to

feed in nearby fields as high tides, lower temperatures, andior rainfall reduce intertidal

foraging habitat and invertebrate prey availability (Goss-Custard 1969, Kelly and

Cogswell 1979, Page et al. 1979, Townshend 1981, Colwell and Dodd 1997). Dunlin

feeding in areas where they perceive the risk of predation to be higher (closer proximity

to predators, landscape features obsûucting lines of sight) respond more vigorously to

alarm calls and move away from those areas, at least ternporarily, following an attack

(Leger and Nelson 1982, Buchanan et al. 1988). 1 related data on movement, macro-

habitat choices, and core area location within the home range to data on prey density and

potential predation risk to indicate whether Dunlin may use knowledge of these factors in

making space use decisions. 1 predict that Dunlin will focus their use of space on areas of

higher food density, and that they will minimize the use of riskier parts of their space

andor modulate their movements while using riskier areas.

Indices of site quality

Previous investigators have found differences in invertebrate prey and Dunlin

densities among sites within the Fraser Delta (MçEwan and Farr 1986, Butler 1992,

Baldwin and Loworn 1994, Butler and Vermeer 1994, Sewell 1996, Sewell and Elner

2001). However, it is unclear if birds that use these different areas differ in performance

(iel the ability to maintain body condition, as an indicator of fitness). 1 used data on prey

densities, predator densities, conspecific densities, a disturbance index, and indices of

13

performance (body condition and percent time spent foraging) to assess the relative

quality to Dunlin of sites within the Delta.

METHODS

Study area

The study took place in the Fraser River Delta (49 05' N, 123 12' W) in south-

western British Columbia (Figure 2.1). The Fraser Delta is the largest estuary on

Canada's Pacific Coast, and is the northernmost site to support a significant wintering

population of the pacifica subspecies of Dunlin (approximately 40,000) (Warnock and

Gill 1996, Shepherd 2001).

The study area included marine intertidal and marsh habitats on Roberts' Bank, in

Boundary Bay, and in Mud Bay, as well as agricultural, urban, and other terrestrial

habitats (Figure 2.1). Dykes separate the marine and terrestrial habitats throughout the

area, and on Roberts' Bank, two jetties extend fiom the shoreline to beyond the low water

mark. The jetties, a ferry terminai (> 3 kms long) and a port facility (> 5 kms long), have

aitered water flow and sediment movement characteristics such that the intertidal

sediments between the jetties are considerably sandier and less penetrable to shorebird

bills than the sediments just north of the port facility jetty (McLaren and Ren 1995, T.

Sutherland, unpubl. data ). Within the study area, median sediment grain size was

highest in Boundary Bay and on Roberts' Bank between the two jetties, and was lower in

Mud Bay and on Roberts' Bank north of the coal port jetty (Sewell 1996, T. Sutherland,

unpuW. data). Mean sediment penetrability decreased with grain size throughout the

study area (Sewell 1996). The tidal regime is such that the intertidal zone on Roberts'

Bank is submerged approximately one hour before and exposed approximsitely one hour

later than the intertidai zone in Boundary and Mud bays (pers. obs., M. Lemon, pers.

comm.).

Capture and marking

Dunlin were caught at three sites in the study area during the winter of 1995196:

Mud Bay (MB), western Boundary Bay (BB), and Westham Island (WI) on Roberts'

Bank (Figure 2.2). Located from east to West through the study area, these sites are

separated by approximately 12 and 9 krns respectively. Dunlin were trapped just outside

the dyke (marine) at MI3 from 20 to 23 December 1995, and just inside the dykes

(terrestrial) at BB and WI from 17 to 23 February 1996. Twenty-five Dunlin were caught

at a fourth site (Brunswick Point) during the winter of 1998 for activity budget sampling.

The activity budget data are discussed in chapter 5.

1 captured Dunlin in mist nets at night and fitted 47 of them (10 at WI, 12 at B8,

and 25 at MB) with 1.45-g radio transmitters (Holohil Systems Ltd., model BD-2G). The

weight of the transmitters represented approximately 3 % of the mean body weight of the

birds, and the best attachrnent location was detennined prior to field work using two

captive Dunlin. Each bird was also fitted with a Canadian Wildlife Service metal band,

and the following measurements were taken to sex them and determine their overall body

size; weight, unflattened wing chord, and culmen length (from the tip to the margin

between mandible and feathers at the center of the upper mandible). Dunlin are sexually

dimorphic, with females being generaily larger than males. A maximum likelihood

model was used to determine sex using culmen length data (Shepherd et al., Chapter 6). 1

put radios on 23 males, 22 females, and two Dunlin of unknown sex, with a 9 1.5%

probability of correctly assigning sex. Juveniles can usually be distinguished from adults

by the presence of bufG-edged inner median coverts until about mid-winter (Page 1974,

Paulson 1993), so Dunlin captured at MB in December were aged using this technique.

Thirteen were adults and 1 1 were juveniles.

Radio telemetry

Sixty-one telemetry stations were initially set throughout the study area (Figure

2.2), and exact Universal Transverse Mercator (UTM) locations (to within 1 2 m) were

recordeci using a GPS unit. GPS accuracy was ascertained using permanent municipatity

monuments (known UTM locations). The range of each radio in the open marine habitat

was tested, and the minimum range was 4 km. The ranges of a subset of radios were

tested in the terrestrial habitat, w h m they were significady lower because of landscape

features and signal interference. 1 could detect radios reliably to a range of 500 m and

fàirly reliably to a range of 1 km. Radio transmitters were detected up to 38 days d e r

deployment. 1 tested telemetry location accuracy by comparing the actual (using GPS)

and estimated (using the telemetry system) locations of three radios hidden randody by

another person in marine and terrestrial habitats within the study area. The mean linear

distance between the actual and estimated locations of the three test radios was 7 1 m 20

m SE.

A dual-Yagi (S-elment) peaWnull van mast telem* system (Wamck &

Takekawa 19%) was used to locate Dunlin within the study area following a three-&y

adjusiment period afler attachent. I drove the van along the dykes and interior roads

From east to West while searching for MB birds, and h m West to east while semhing for

Wi and BB birds. Two compass beryings were taken on each Dunlin frorn consecutive

telemetry stations, and the time between bearings was minimized ( m m = 6.4 minutes). I

stopped at a given telemetry station and listened for al1 the radios that had not as yet been

located, making a list of al1 signals heard. 1 then took bearings on the radio signals 1

heard, &ove to the next telemetry station, and took a second bearing on each of those

radios before once again listening for the radios that had not yet ken located. If the radio

signal was weak and appeared to corne fiom fariher away than the next station, 1 collected

another set of bearings closer to the radio signal in order to increase the accuracy of the

location. Otherwise, once 1 had collected a location for a given radio, 1 deleted that radio

h m the receiver memory and did not listen for it again until the next telemetry m. If 1

heard a number of radios in the same area, 1 drove back and forth between the two

telemetry stations collecting bearings on a few radios at a time in order to minirnize the

t h e lapse between bearings. During high tides when 1 could visually locate flocks near

the dyke that included radio-marked Dunlin, 1 collected locations by meairring the

distance along the dyke between the flock and the nearest telemetry station using the van

odometer.

Dunli in the Fraser Delta are active both day and night, and tirne of day has been

shown to affkct Dunlin habitat use patterns (Mouritsen 1994). 1 therefore collected

location data &y and night during standardized ûacking runs f: 3 hours h m high and

low tides. Daytime was considered to be one half how before to one half how afier

sunrise and sunset. Activity sampling also took place throughout the 24-hour day and

across tide stages, by listening for activity-switch mediated changes in radio pulse rates

when a bird changed posture fiom head-up to head-down. 1 record4 pulse rates and

general locations for each radio-marked bird within hearing range every 15 minutes, and

precise locations of the birds king sampled were recorded using the van-mast telemetry

system approximately every two hours. This process is described in more detail in

chapter 5. In recording the general location, 1 noted whether the bird had made any

lateral movements (paralle1 to the shore) in the 15 minutes since the last pulse rate

reading. In the marine macro-habitat, 1 did not include movements with the rising or

ebbing tides that ran perpendicular to the shoreline. 1 categorized these movements as

small-scale (more than approximately 100 m but less than approximately 1 km), and

large-scale (more than approximately 1 km). 1 located MB Dunlin during 42 ûacking

runs, and during 67 hours of activity sampling, covering al1 but two days h m 26

December 1995 to 26 January 1996 (the first tracking period). 1 located WI and BB

Dunlin during 45 backing runs, and during 64 hours of activity sampling, on al1 but 6

days from 21 February 19% to 28 March 1996 (the second ûacking period).

1 did not swvey the entire study area on each ûacking nin due to tirne constraints,

but since Dunlin were rarely located far fiom their banding sites, it is dikely that the

partial surveys resulted in many missed Dunlin locations. Only 1.1 % (N = 5) of the 440

locations of MB Dunlin occurred on Roberts Bank, even though Roberts Bank was

surveyed for MB birds on 67% (N = 28) of the ûacking runs. Only 3.6% (N = 16) of the

441 WI and BB locations occurred in the area associated with MB, even though the ME3

area was surveyed for WI and BB birds on 53% (N = 24) of the tracking rus. The inland

tracking stations between WI and BB (Figure 2.2) were only used during the second

tracking perd. They were added after data collected during the first tracking period

revealed significant use of upland areas by Dunlin at night. Most of the radio-marked

17

Dunlin using upland areas around Ml3 could be located h m the dykes. This was not the

case for the extensive upland area between WI and BB, so additional tracking stations

were added. Twenty-four more stations were added for time-activity budget sampling in

1998 (Chapter 5) (Figure 2.2).

1 used only one location per bird per telemetry rwi (in other words, per tide stage)

in the analyses. The locations were considered to be statistically independent since tidal

action aitered habitat availability between subsequent high and low tides and the time

between tracking runs was sufficient for D d i n to fly to any point in the Delta. An

Arcview program (McCullough 1996) was used to triangulate the bearings and obtain

UTM coordinates for each individuai Dunlin located on each tracking m.

Marine invertebrates

To assess marine invertebrate density among sites and intertidai micro-habitats, 1

collected samples fiom each micro-habitat dong transects near each of the three banding

sites. Sarnples were collected during the p e n d between the two Duniin tracking bouts. 1

used data on sediment grain size and marine plant species distribution fiom several

sources (Swinbanks 1979, Fisheries and Environment Canada 1977, Dunn et ai. 1993,

McLaren and Ren 1995) to digitize a base map of the study area that characterized and

delineated micro-habitats w i t h the intertidal zone. Transects were set perpendicular to

the shoreline and extended out 1.6 km into Mud Bay, 2.1 km into Boundary Bay, and 4.1

km on Roberts' Bank so that they would pass through each micro-habitat type at each site

(Figure 2.3). There were three micro-habitat types on Roberts' Bank (sandy mud RB,

muddy sand RB, and sand RB), three in Boundary Bay (algai matlsand, sand BB, and

eelgrass/sand), and two in Mud Bay (sandy mud MB and muddy sand MB) (Figure 2.3).

Roberts Bank receives considerably greater freshwater and other inputs than Boundary or

Mud Bays, so aithough both areas contained sandy mud, muddy sand, and sand micro-

habitats, 1 assumed there would be differences in invertebrate density between the areas

and treated them as separate micro-habitats (hence the identifying initiais). Invertebrate

samples were collected during the day under clear weather conditions h m each micro-

habitat dong each of the three transects.

Sewell and Elner (2001) studied the spatial scale of variability in invertebrate

abundance of Bounàary Bay and found signifiant differences at scales of 10's and 100's

of meters, but not at the scale of 1 km. Thecefore, 1 collected three random cores dong

the edge of a 1 m diameter circle at each of 3 random sites dong the edge of a 100 m

diameter circle radiating from a point approximately at the centre of each micro-habitat

type. C. a. pac$ca bill lengths are between 3.3 and 4.5 cm long (Shepherd et ai.,

Chapter 6), so the cores were 5 cm deep (assumed to be the maximum probing depth) and

10 cm in diameter. Some invertebrates may have retreated to below the core depth upon

disturbance to the sediment, although, due to the small size of most of the invertebrate

species (large annelids were those > 1 cm long) the effect may not have been significant.

In terms of the statistical comparisons made, al1 cores were collected in a standardized

fashion and 1 assumed that the likelihood of invertebrates escaping was similar among

coces. Cores were fiozen and later sieved with a 500-micron sieve to retain al1 macro-

faunal invertebrates, and ail retained invertebrates were placed in vials containing 95%

ethanol. Dunlin may be capable of consurning meiofaunal invertebrates that would have

passed through the 500-micron sieve (Zwarts et ai. 1990, Suthedand et al. 2000),

however we measured only macro-faunai prey in this study. The invertebrates were

counted and separated into the following categories under a dissecting microscope: large

annelids (11 cm in length), small annelids (4 cm in length), crustaceans, and molluscs.

Invertebrates h m each of these categories had been found in Dunlin guts in the Fraser

Delta and in neighbouring Washington State (Brennan et ai. 1990, B. Elner, pers.

comm.).

Rainfail cm affect shorebird feeding rates in intertidal habitats through a decrease

in prey detectability and/or by altering prey behaviour (Goss-Custard 1970b, Pienkwoski

198 1, Goss-Custard 1984). Warnock et al. (1 995) found a significant negative correlation

between the numbers of Dunlin wintenng on Bolinas Lagoon and the amount of local

raiafall, and suggested the phenomenon could be due in part to the effects of rainfail on

the burrowing depth of marine invertebrate prey. In order to determine whether rainfall

affected the accessibility of invertebrates in the top 5 cm of the sediment, 1 made a second

collection (as described above) dong the Mud Bay transect during a pend of daytime

19

rainfall. Intertidal invertebtates can also move closer to the sediment surface and become

active at night (Dugan 198 1, Evans 1987, Robert and McNeil 1988, McNeil et al. 1999,

so 1 made a third collection (as described above) dong the Mud Bay transect under clear

weather conditions at night.

STATISTICAL ANALYSES

All statistical test results were considered to be significant at P < 0.05, however,

resuhs with P-values between 0.05 and 0.1 are reported as possibly significant

biologically. Interaction terms were considered to be statistically significant at P < 0.10,

since significance tests for interaction terms have lower power than those for main effects

(Littell et al. 1991). Results of 2 or 3-way ANOVAs reported here were those of the

reduced models in cases where the interaction tenn was not significant, and 1 report least-

squares means taking the other factor(s) into account.

Regional site fidelity

1 determineci the total nurnber of days chat each radio-marked Dunlin remaining in

the study area failed to be Iocated, as well as the length of any absences. To minimize the

likelihood of classiQing birds as "absent" when they were simply out of range, 1 assessed

regional site fidelity on a daily basis rather than by tracking m. In general, individuais

were tracked more than once per 24-hour day, so to be classified as "absent", W i n

would have to fail to be located two or more times in a given 24-hour day. 1 also

calculated the average percentage of ûacking runs, by tide stage and tirne of day, ihat

individual Dunlin failed to be located in the study area (using only tracking runs that

covered the entire study ma).

Local site fideliîy

1 examined the locations of each bird that remained in the study area for at least

two weeks (n=39). 1 calculated the numbers and proportions of radio-marked Dunlin, and

the overall proportions of Dunlin locations, that occurred outside of the areas associateci

with the birds' banding sites. There was little home range overlap among Dudin from

the three banding sites, so 1 used landmarks near the limits of these non-overlapping areas

to mark off the areas associateci with each banding site. The coal port jetty and upland

road marked the boundary between WI and BB areas, while 88' Street (and the imaginary

line following it out to the low water mark) marked the boundary separating the BB and

MB areas (Figure 2.1).

Individual home range and core area estimates

1 estimated the home ranges of radio-marked individual Dunlin to determine how

they distributed themselves in the Fraser River Delta and whether they showed within-

year fidelity to local areas. 1 estimated fixed kernel95% and 30% utilization distributions

(UDs) to represent Dunlin home ranges and core use areas respectively. The 95% UD is

the one most often used to represent an animal's home range. 1 selected the 30% UD, the

area within which Dunlin spent approximately one third of their time, to represent the

core area. The home range and core use areas are therefore the areas within which an

individual Dunlin has a 95% or 30% probability (respectively) of king located. Cote

areas are dehed as "areas of concentrated use within the home range" (Samuel et al.

1985). The Dunlin in my sample spent 30% of their tirne in the area within their 30%

UD, but that area represented on average only 7.6% + 1.3 SE (range = 3.6-16.9%) of the

mean size of the 95% UD. Therefore the 30% UDs of the Dunlin in my sample meet the

criterion for classification as core areas.

My data did not conform to a bivariate normal distribution, so 1 used a non-

parametric kernel density estimator, and 1 chose the fixed kernel method, as it provides

the least-biased estimates of home range (Worton 1995, Seaman et al. 1999). 1 estimated

home ranges and core areas using the Arcview GIS and Spatial Analyst programs and the

Animal Movement Analysis Extension (Hooge & Eichenlab 1997), and 1 used least

squares cross validation as the smoothing parameter.

1 calculated UDs for each Dunlin with 15 or more radio locations (3 1 birds).

Locations collected during activity budget sampling (one per tide stage) (Chapter 5) were

included in analyses where they did not overlap temporally with locations collected on

standardized tracking nuis. 1 comparai home range and COR area sizes inclding and

excluding activity locations separately by site using ANOVA, and found no differences

(F ,,,,, c0.8, P >0.31 forall sites).

Due to the short battery life of the srnall transmitters (limitai by the size of my

study species), 1 obtained the 30 or more statistically independent locations recommended

by Seaman et al. (1999) for only a few individuals. 1 therefore performed three tests to

determine whether location sarnple size influenced the s ix or shape of the home ranges

in this study. 1 log-transfomed the data and performed eaçh test separately by site, since

there was a significant difference in home range size among sites (sec below).

1 used ANOVAs to determine whether home ranges ancilor core areas estimated

using the smallest (< 20) or largest (> 30) numbers of locations differed in size h m

those of the rernaining birds. Three Dunlin had to be removed to achieve non-

significance (P > 0.32 for al1 sites), but two of these were restored to the data sets since

the size of their home ranges and core areas did not appear to be due to %nusual erratic

wanderings" (censu Brown 1975). Next, 1 used correlation analysis to detemine whether

fixed kemd home ranges and core areas varied with sample size. One more Dunlin had

to be removed to achieve a non-significant relationship (P > 0.1 for al1 sites), but it was

restored to the data set for the reason outlined above. Finally, 1 used correlation analysis

to determine whether the outer (95%) and inner (30%) dimensions of each bird's home

range were proportionate, and whether these were proportionate to middle dimensions

(50% and 75% UDs). Al1 üDs were significantly positively correlateci (r > 0.83, P <

0.001 for al1 sites and tests). Thuty of the original 3 1 Dunlin remained in the data set

once the sample size tests were completed.

Randomizption models for anaiyses using individual Dunlin

My samples of individual radio-mark4 Dunlin ofien did not meet the assumption

of normality, so 1 used randoniuation models to test the robustness of the results of ail

ANOVA and regression analyses describeci in this thesis. For example, to randomize the

results of a 2-way ANOVA testing for sex and site differences in home range size, 1

randomly shuffled the sex and site designations of the birds in the data set and ran

ANOVAs on 1000 samples generated in this way. 1 then obtained a P-value by

comparing the F-statistic obtained h m the ANOVA using the actual sex and site

assignments to the distribution of F-values obtained fiom the raadomly generated

samples.

Individual home range and core area comparisons

1 used 2-way ANOVAs and Bonferroni adjusted multiple t-tests on log

transformed data to compare home range and core area sizes among categories of Dunlin

(sex and site, and sex and age at the MB site). On average, appmximately two thirds of

the habitat within the average Dunlin's home range and core area was marine, while the

remaining third was terrestrial. Factors such as food resources and predation risk, that

may affect home range and core area sizes, differ in nature between the marine and

terrestriai habitats. 1 therefore repeated the analyses outlined above separately on the

marine and terrestrial components of the Dunlins' home ranges and core areas. 1 also

used correlation ansilysis to determine whether the marine and terrestrial components

were positively correlated with each other.

Individual macro-habitat choices within home ranges

To characterize the use of marine versus terrestrial macro-habitats within each

Dunlin's home range, 1 calculated the percentage of locations in the terrestrial habitat by

time of day and tide stage for each individuai Dunlin. Radio-marked Dunlin were rarely

located in the terrestrial habitat during the day (only 1.6% of al1 the day-time locations)

(Figure 2.4), while 46.9 % of the night-time locations occurred there (Figure 2.5). 1

therefore used only night-the data to make comparisons of the percentage of locations in

terrestrial habitat among sex, age, and site categories of Dunlin. Since within-individuai

radio location sample sizes varied, 1 weighted the percentages by the nurnber of locations

obtained. There was a significantly higher percentage of Dunlin locations in the

terrestrial habitat during hi& tide (77.2 % 6.9 SE) than low tide (1 5.2 % 5 4.1) (F,,, =

98.2, P < 0.001), so tide stage was included as a factor in al1 comparisons among

categories. 1 made the comparisons using 2- and 3-way repeated measwes ANOVAs for

23

unbalanced samples, and used Bonferroni adjusted multiple t-tests for site comparisons. 1

also regressed the percent of locations in terrestrial habitat on culmen length, since one of

the possible reasons for use of terrestrial habitat might be a decrease in the availability of

marine invertebrate ptey due to their burrowing deeper into the sediment column.

Most Duniin were absent fiom the study area during the occasional tracking nui.

1 calculated the proportion of day and night high and low tide tracking runs that each bird

failed to be located in the study area (using only tracking runs that covered the entire

study area) as part of the regional site fidelity analysis. 1 then used Kruskal-Wallis and

Wilcoxon signed rank tests to make comparisons arnong site, sex, and age categories of

Dunlin to assist the interpretation of the macro-habitat choice results.

Individual movements within home ranges

To assess in greater detail how Dunlin used the area within their home ranges, 1

calculated the within-bird mean distance moved between consecutive hi& and low tides,

weighted by sample sizes. 1 made comparisons among categories of Dunlin (sex and site,

and sex and age at the MB site) using 2-way ANOVAs and Bonferroni adjusted multiple

t-tests on log-transfonned data. Dunlin included terrestrial as well as marine habitats

within their home ranges, but their use varied by time of day. 1 therefore made movement

comparisons by time of day (within day, within night, between day and night) and by

macro-habitat (within marine, within terrestrial, between marine and terrestrial) using

repeated measures ANOVAs for unbalanced samples.

1 also collected data on the fkquency of lateral movements as an index of

disturbance, since Duniin often move/fly in response to predators or otlier disturbances

(ûierschke 1998, pers. obs.). 1 calculated the percent of 15-minute time blocks SE)

that individual Duniin made small-scale (0.1-1 .O km) and large-scale (> 1 km) lateral

movements. 1 made within-bird comparisons by macro-habitat and time of day using

one-way repeated measures ANOVAs, and arnong categories of Dunlin (sex and site, and

sex and age at the MB site) using Zway ANOVAs and Bonferroni adjusted multiple t-

tests. Dudin were abject to predation throughout the 24-hour day by both falcon

(diunial) and owl (diurnal and noc tud) species, however, falcon species appeared to be

24

their primary predators (Mortality section, Chapter 5). 1 therefore compared movement

frequency among categories of Dunlin using two separate data sets; one c o v e ~ g the

entire 24-hour day, and one covering daytime only.

Marine food resources

1 calculated the mean densities pet m2 of four marine invertebrate types that have

been recorded in the diet of Dunlin in Washington state and the Fraser River Delta

(Brennan et al. 1990, B. Elner, pers. cornrn.): large annelids, small annelids, crustaceans,

and molluscs, within each micro-habitat at each site. 1 created a twodimensional map of

available invertebrates by applying the mean densities to the base map of intertidal micro-

habitats (Figure 2.3). This involved some lateral extrapolation of values across the

intertidal zone. However, Sewell and Elner (2001) examined the spatial scale of

variability in the abundance of invertebrates in Boundary Bay and found significant

differences at scales of 10's and 100's of meters (incorporated into the sarnpling scheme),

but not at the scale of 1 km or "sides of the bay" (3.5 km) for polychaetes, Corophium

species (crustacean), or total individuals.

1 dso calculated the mean densities of each invertebrate type during a dry day, a

dry night, and a wet day within each micro-habitat dong the Mud Bay transect. 1 used

separate paired t-tests to compare invertebrate densities between day and night (nvo-railed

test) and to determine if densities were lower during times of rainfall within each micro-

habitat (one-tailed test).

Individual use of space (marine) in relation to prey density

1 layered the marine home ranges and core areas of each Dunlin over the

invertebrate base map described above. 1 used only the marine component of the home

ranges and core areas for al1 of the analyses relating to the distribution of marine food

resources. 1 calculated the overall mean densities of each of the four invertebrate types

occurring within the marine areas used by each bird.

1 used 2-way MANOVAS to compare the densities of the four invertebrate types

within the marine home ranges and core areas of different categories of Dunlin (sex and

site, and sex and age at the h4B site). 1 also tested each invertebrate grpe separately using

2-way ANOVAs, and used paired t-tests to determine whether there were higher densities

of invertebrates within a Dunlin's core area than within its home range.

Indices of site quality

1 assessed the basic suitability of each of the three sites in the Delta using data on

relative prey densities (see above), frequencies of movement as an index of disturbance

(see above), and avian predator nurnbers (see below) (Fretwell 1972). 1 then incorporated

data on conspecific densities and performance indices to detemine whether individual

Dunlin experienced particufar sites as king of greater or lesser qualiiy than others.

Both falcons and owls are known predators of Pacifica Duniin (Page and

Whiteacre 1975, Kus 1 985, Dekker 1998 and 1999, personai observation), and remains of

radio-marked Dunlin from this study were collected fiom 3 Peregrine Falcon (Fulco

peregrinus) roasts and 1 Barn Owl (Tyto alba) roost (Chapter 5). 1 detmined the

numbers of Dunlin and their predators in the areas associated with each banding site

(described above in the "local site fidelity" section) using data collected during Christmas

Bird Counts fiom 1995 through 1998 (Ladner survey areas D, F, G, H, and 1 (Jude Grass,

unpubl. raw data)). 1 calculated a density estimnte for the Dunlin by dividing their

numbers by the areas of mudflat (at mean low water) associated with each site.

1 used data on body size, body weight, and mean percent time spent foraging to

give a site-specific index of Dunlin performance. To obtain a measure of body size, 1

performed a principal component analysis with data on culmen length, wing length, and

weight (Rising and Somers 1989). The first principal component eigenvectors had

similar signs and magnitudes for each measure (0.56-0.59), and ihe cumulative

correlation matrix eigenvalue was 0.68, so PC 1 was used as my measure of body size. 1

did not use birds h m the MB site in this anaiysis, since foraging activity was

significantly affected by season (Chapter 9, and MB Dunlin were tracked at a different

time of year than BE and WI Duniin. 1 used one-way ANOVA to compare PC 1 between

WI and BB and found no ciifference (F,,, < 1.9, P > 0.25). 1 then used one-way ANOVA

to compare body weight between WI and BB as an index of condition. 1 used two way

26

ANOVAs with tide stage to test for differences in the percent of t h e spent foraging

between the two sites. Al1 analyses were done separately for each sexy since Dunlin

females are larger and the sex ratios at the two sites were not exactly the same.

RESULTS

Regional site fidelity

Thirty-nine of the 47 radio-marked Dunlin survived and remained in the study

area for at least two weeks. Of these, 19 were located in the study area every &y 1

tracked hem, and 9 were missing on one day only. Two Dunlin were absent for more

than four days, one of which was located approximately 100 kms south of the Delta in La

Conner, WA, before it retumed 1 1 days later. These two birds were therefore dropped

from the analysis of local site fidelity. Neither of these longer absences was associated

with a period of inclement weather. However, during a two-week freeze toward the end

of the tracking period, when fields and mudflats were covered by ice and snow, al1 of the

radio signals in the Delta disappeared. 1 cannot say for certain whether al1 of the birds

lefi or whether some of the radios simply reached the end of their battery life, but 1 also

obsewed a decrease in the size of the Dunlin population (from approximately 20,000 to

none) at that tirne. Within two days of the start of the fieeze, the number of active radios

dropped from 13 to 7, then down to 3 two days later, and down to zero three days d e r

that. If the birds had died of starvation in the Delta, 1 would have found the remains

(Chapter 5). An aerial reconnaissance between Vancouver and Seattle, WA, undertaken

at the start of the keze, detected no radio signals south of the Fraser Delta. However,

the freeze extended geographically beyond my reconnaissance range.

Dunlin can be site faithful between as well as withii years (Warnock 1994). One

Dunlin that cacried a radio in spring 1998 was killed hvo non-breeding seasons later by an

overhead wire less than 500 m h m its original banding site. Since 1 only radio-marked

25 Dunlin in 1998, out of the approximately 40,000 birds that spend the non-breeding

season in the Fraser Delta (Shepherd 2001), the likelihood of obtaining even a single

between-year data point is extremely low.

Most Dunlin were absent h m the study area during the occasional tracking m.

During the &y, the average Dunlin was absent fiom 9.3% & 2.1% SE) of high tide

tracking runs, and 16.6% (t 3.6% SE) of low tide tracking runs. At night, the average

Dunlin was absent from 36.5% 5.7% SE) of high tide tracking runs, and 21.6% (lr

4.9% SE) of low tide tracking runs. Some of these absences may have been due to the

birds having moved out of range of the telemetry system (at least 0.5 to 1 km inland or

greater than 4 km offshore). At night, in particular, Dunlin were more likely to use

upland sites and low tides were lower than they were during the day. During daytime

high tides, some of the absences would also have been due to Dunlin flying in flocks fac

out over the water, likely a tactic to avoid predation (Breman et al. 1985, Dekker 1998).

Local Site Fidelity

Duniin fiom each banding site showed a high degree of within-season fidelity to

that site. There was much home range overlap arnong individual Dunlin within sites, but

there was little overlap among Duniin fiom the three banding sites (Figures 2.6-2.8). One

of seven Duniin fiom WI was located once in Boundary Bay and once between the two

jetties, and a second Duniin h m WI was located once near Mud Bay. Al1 of the

remaining locations of Wl Duniin (98.2%) occurred north of the coai port jetty. Of nine

BB birds, two were only ever Iucated north of the coal port jetty, one was located north of

the coai port jetty on one occasion, and another was located there twice. The remaining

five were oniy located south and West of the coal port jetty. Oniy 6.0% of the BB Duniin

locations were east of 88' street. Of 23 MB birds, 17 were never located West of 88"

street, and one was oniy ever located West of 88" street. Aside h m that bird, just 6.5%

of Mi3 Duniin locations occurred West of 8gh Street and only 2.1% West of the BB airport

(Figure 2.1).

Individual home range and corn areri cornparisons

Results of the separate analyses of the marine and terrestrial components of

Dunlin home ranges and core areas were consistent with those of the total home ranges

(Tables 2.1 to 2.3). Marine and terrestrial home range components were also positively

28

correlated with each other (r = 0.91, P < 0.001), while core areas were not correlated with

each other (r = 0.17, P = 0.41).

Duniin in BB had larger home ranges than Dunlin at WI and MB, but WI and Mi3

Duniin did not differ fiom each other (Tables 2.1 and 2.2, Figures 2.6-2.8). Males had

smaller marine home ranges than females (Tables 2.1 and 2.2), and there was some

indication that male overall home ranges may also have been smaller than those of

females = 4.3, P = 0.05). The interaction between sex and site was significant for

terrestrial home ranges, so 1 tested for site differences within sexes and sex differences

within sites. BB Dunlin had larger terrestrial home ranges than the other two sites for

both sexes (F,,,,, > 6.0, P < 0.02 for both tests). However, sex differences occurred oniy

at the MB site (FI.,, = 10.4, P = 0.006), where male terrestrial home ranges (5.7 km2+ 0.9

SE) were smaller than those of females (10.7 km2 + 0.9 SE). Home range size did not

differ between age categones (Tables 2.1 and 2.2).

Again, BB Dunlin exhibited the largest core areas overall and by macro-habitat

(Tables 2.1 and 2.3). There were no differences in core area sizes between sex or age

categories of Dunlin for the terrestrial component (Tables 2.1 and 2.3). There were

significant interactions between age and sex categories in the sizes of overail and marine

core areas at MB. 1 therefore tested forage differences within sex classes and sex

differences within age classes. The oniy significant rcsult was that juvenile females had

larger overail core areas (1.5 km2 + 0.4 SE) than juvenile males (1.2 km' 0.1 SE) (F,, =

7.2, P = 0.02).

Individual macro-habitat choices witbin home ranges

Most (> 70?!), but not dl, of the Duniin wintenng in the Fraser Delta exhibited

some switching between marine and terrestrial macro-habitats, primarily at night, and

during hi& tides. Radio-marked Dunlin were rarely located in the terrestrial habitat

during the &y (only 1.6% of al1 the day-the locations) (Figure 2.4), while 46.9 % of the

night-tirne locations occurred there (Figure 2.5). Dunlin spent a significantly higher

percentage of t h e in the terrestrial habitat during high (77.2 % f 6.9 SE) than tow tide

(1 5.2 % + 4.1) (FI,, = 98.2, P < 0.00 1). In addition, there was considerable individual

variation in the use of the terrestrial habitat at night, with some Dunlin never located

there, and some aiways located there (Chapter 2).

There was some indication that the terrestrial habitat may have been used more at

BB than at the other two sites (P = 0.06, Table 2.4). Males were more likely to be located

in the terrestriai habitat (64.4 % + 5.8 of locations) than females (43.3 % + 7.3) (F,,, =

5.1, P = 0.03). However, inter-sexual dierences were reduced when tide stage was

considered (males = 61.5 % + 5.3, and females = 46.4 % + 5.3), and the difference was no

longer significant (F,,, = 4.0, P = 0.1 1). There was some indication that there may have

been a negative relationship between culmen length, a continuous measwe of sex ratio,

and the percent of locations in terrestrial habitat (FI,, = 3.7, $ = 0.10, P = 0.06). There

were no differences among sites (P > 0.25 for al1 tests), or between sexes (P > 0.29 for ail

tests) in the percentage of day or night high or tow tracking runs that Dunlin failed to be

located in the study area.

There was a significant interaction between age, sex, and tide stage in the

percentage of time MB Dunlin spent in the terrestrial habitat (FI,,, = 8.0, P = 0.06).

Within MB, there were no differences between age classes in the percentage of tracking

nuis that Dunlin failed to ùe located in the study area (P > 0.25 for al1 tests). Nor was

there any difference between the sexes for the low tide and day-time high tide tracking

runs (P > 0.55 for ail tests). However, Mi3 males were absent fiom 25.0 % + 5.8 SE of

night-time high tide tracking runs, while females were absent h m 56.3 % + 10.3 SE

(WSRT, P = 0.03). Since the direction of the sex difference in absence matched the

direction of the difference in macro-habitat choice, the high tide data were dropped and i

used a 2-way ANOVA to test for age and sex differences in macro-habitat choice during

low tide only. There was no age effect (F,,,, = 0.5, P = 0.51), but there was some

indication that there may bave been a sex effect (FISI, = 4.3, P = 0.06).

Indmdual movements within home nages

There was a significant interaction between site and sex categories when

comparing the mean witbin-bird distance moved between consecutive high and low tides

(Fu, = 6.9, P = 0.01). Thecefore 1 tested for site differences within each sex category,

30

and sex differences within each site category. There was no site effect within females

(Table 2.5). Male movement distance increased with home range size among sites, but

only BB and Mi3 males difXered significantly by Bonferroni adjusted multiple t-tests

(Table 2.5). The only significant sex effect was that BB males moved greater distances

than females (Table 2.5). There was also a significant interaction between age and sex

categories for the MB birds (F,,,, = 4.1, P = 0.08), so 1 tested for age differences within

each sex category, and sex differences within each age category. None of the

comparisons were significant (F,,, = 0.03-6.1, P 2 0.09 for al1 tests),

Dunlin moved longer distances by day (3.2 km 2 0.5 SE), and between day and

night (3.3 km + 0.3), than by night (1.5 km + 0.4) (F,, = 7.2, P = 0.001). They also

moved farther in the marine habitat (3.2 km f 0.4 SE), and between marine and terrestrial

habitais (3.4 km + OS), than within the terrestrial habitat (0.7 km + 0.4) (F,,, = 16.0, P <

0.001).

The kquency of large-scde movements differed only by macro-habitat (Table

2.6), with Dunlin moving more ofien in marine than in terrestrial habitat. Dunlin also

made more small-scaie movements in marine than in terrestrial habitat, and during the

day than at night (Table 2.6). During daytime, there were no age or sex effects on the

frequency of small-scale movements (F,,,, < 1.3, P > 0.23). Throughout the 24-hour day,

there was no age effect on small-scale movement frequency, but females moved more

oflen than males (Table 2.7). For site comparisons, see Indices of site quality section

below.

Marine food resources

Mean densities per m2 of each marine invertebrate type (large annelids, smail

annelids, cnistaceans, and molluscs) within each micro-habitat at each site are reported in

the appendix, Table 2.A. 1. Smail annelids were less available to Dualin when it was

raining than when it was dry (Appendix, Table 2.A.2). At night, there was some

indication that large annelids may have been more available and that crustaceans rnay

have been less available than they were during the day (Appendix, Table 2.A.3).

tndividual use of space (marine) in relation to prey density

Dunlin h m BB, where marine home ranges were the largest, had the lowest

densities of crustaceans and molluscs and intermediate densities of large and small

annelids within them (Tables 2.8 and 2.9). With ail 4 classes of invertebrates taken

together (collectively), there was no difference between the sexes in the densities of

invertebrates found within Dunlin marine home ranges (Table 2.8). Individually, there

was some indication that the marine home ranges of males, which were smaller than

those of fernales, may have contained higher densities of crustaceans than did the

fernales' ranges (P = 0.06, Tables 2.8 and 2.9). In fact, although the differences were not

statistically significant, male marine home ranges contained higher densities of ail four

invertebrate types than female marine home ranges. There were no age differences, either

collectively or individually, in the densities of invertebrates found within Dunlin home

ranges (Tables 2.8 and 2.9).

The results for core areas parallel those for home ranges. Again, Dunlin from the

site with the largest marine core areas (BB) had the lowest densities of crustaceans, large

annelids, and small annelids within them (Tables 2.8 and 2.9). Collectively, there was no

difference in core area size between the sexes, however, individually, the core areas of

males contained more crustaceans per m' than did those of femaies. There were no

differences between age classes, either collectively or individually, in the densities of

invertebrates found within Dunlin core areas (Tables 2.8 and 2.9).

Within individuals, there were higher densities of crustaceans and smail annelids

contained in their core areas than in their home ranges (paired t-tests, Table 2.10).

Indices of site quality

lnvertebrate prey densities were generaily lowest at BB and similar between WI

and MB. Dunlin home range size, which was closely related to invertebrate density

(Chapter 3), increased h m Wi to MB to BB. The density of Dunlin was lowest at BB

(1 OUkmZ), intermediate at MB (1 SOikm3, and highest at WI (1 9Ukm3, and the nirmbers

of fdcons were highest at WI and lowest at Ml3 (Table 2.1 1).

During daytime, smail-scale movement îrequency decreased fiom WI (26.4 2

3.0), to MB Dunlin(21.5 + 1.5), to BB (16.3 4 2.5) (FIJz = 4.2, P = 0.009, WI and BB

diKerent by Bonferroni adjusted multiple t-tests). Throughout the 24-hour day, Dunlin

fiom WI made more frequent srnail-sale movements than Dunlin from either BB or MB

(Table 2.7).

To determine whether there was likely to be any difference in performance

between Dunlin fiom BB and WI, 1 compared their body weights (after determining that

there was no difference in body size between sitcs) and percentages of time spent

foraging. Thete were no differences between WI and BB Dunlin in body weight or

percent time spent foraging within each sex (Table 2.12).

DISCUSSION

Individual Dunlin wintering in the Fraser River Delta were site-faiffil and

appeared to modulate their use of space in relation to prey density and predation risk.

Duniin marine home range and core area sizes were largest at the BB site, where overall

prey density within them was lowest. Dunlin core areas contained higher densities of

crustaceans and small annelids than did their home ranges, indicating that Dunlin

focussed their use of space in the better feeding areas within their ranges. Dunlin made

little use of the higher-cisk terrestrial macro-habitat during the day, when their main

predators (falcon species, see below) were active, but made considerable use of it at night.

Duniin appeared to modulate marine home range size in relation to marine prey

density. It is possible that such a relationship could have acisen by chance if larger home

ranges included more low-preydensity habitat by nature of king larger. In order for this

to be a viable alternative explanation, there would have to be a greater proportion of low-

prey-density than high-prey-density habitat available to Dunlin. In the Fraser River

Delta, mollusc density exceeded the median density in 64.1% of the available marine

intertidal habitat, crustacean density exceeded the median density in 42.1% of the

available habitat, small annelid density exceeded the median density in 35.9% of the

available habitat, and large annelid density exceeded the median density in 48.3% of the

33

available habitat (Appendix 2.1). Had the median density of cmstaceans been 1.5%

lower, then crustacean density would have exceeded the median density in 69.6% of the

available marine intertidal habitat (Appendix 2.1). It is therefore unlikely that Dunlin in

the Delta using larger home ranges hcluded significantly more low-preydensity habitat

than Duniin using smailer home ranges simply by chance.

Site îidelity

Duniin in the Fraser Delta showed a high degree of within-year site fidelity, both

regionally and locally, and cm be site faitfil between years as well, as evidenced by the

return of an individual radio-marked two winters earlier to within 500 m of its original

banding site. Invertebrate prey resources appeared to be plentifid in the Delta (Appendix,

Table 2.A.1, McEwan and Farr 1986), and since Dunlin apparently used knowledge of

the distribution of invertebrates to regulate their use of space, they can be expected to

benefit from being site-faithful. Alternative sites may offer less productive habitat andfor

more competition, especiaily with influxes of Dunlin from the sites afîected by severe

weather.

In January 1996, when ice covered the foraging habitat, Dunlin moved away fiom

the estuary, but their nurnbers retumed to near pre-keze levels shortly after melt. 1

counted 20,200 Dunlin in Boundary Bay and Mud Bay on 18 January 1996. The ûeeze

lasted h m 21 January to 3 Febniary 1996, during which time the population dropped to

zero. The number of active radio signals dropped fiom 13 on 19 January to three on 24

January (three days d e r the start of the freeze). Radio-marked Dunlin were last heard in

the Delta on 26 January. This decline in the number of radio-marked birds was likely due

to their having left the area. Had these buds died in the Delta, 1 would have been able to

use their radio signals to locate their carcasses. Alternatively, since the fieeze occurred

near the expected end of the radio transmitters' battery lives, some of the radio signais

codd sirnply have ceased to transmit. AAer melt, on 7 February 1996,I counted 17,600

Duniin dong the same survey route. Since the fieeze lasted for two weeks, by which

tirne ail of the radio batteries would have died, no radio-marked Dunlin were detected

again when population numbers retwned to pre-kze levels &er the melt. 1 therefore

34

cannot be certain that the Dunlin that returned to the Delta were the same individuals that

had been there before the freeze, but this is the most parsirnonious explanation for the

shilarity in pre- and pst-freeze Duniin population numbers. The lower number of

Dunlin post-fieeze may reflect mortality ancilor birds that chose not to retum after the

melt.

Locally, there was very little home range overlap among Dunlin h m different

sites. Birds that foraged in areas of sandy sediment were less likely to be located in areas

of muddy sediment and vice versa. Rands and Barkham (198 1) and Gemtsen and van

Heezick (1985) found that Dunlin feeding tactics differed between sandy and muddy

substrates, and experience can improve foraging efficiency (Groves 1978, Goss-Custard

and Durell 1987, Caldow et al. 1999). Foraging almost exclusively in one substrate or

another may aHow individuals to specialize in a particular foraging technique and thereby

improve their food intake efficiency.

Home range and core area size

1 collected data fiom the MB site in winter (Dec-Jan) and from the Wi and BB

sites in spring (Feb-Mar), but 1 have no reason to believe that the differences between

sites are due to seasonal changes. Warnock and Takekawa (1996) did not find that season

afTected the home range size of Western Sandpipers wintering in California In my study,

the mean home range size of the MB Dunlin was intermediate to those of the WI and BB

Dunlin, and there was greater variation between the two spring sites than between the

winter site and either of the spring sites. 1 therefore pooled al1 of the data and discuss

home range and core area size differences in relation to banding site rather than to season.

Marine home ranges and core areas were largest at BB, where overall prey density

was lowest. The larger mean grain size at BB may also decrease the availability of

minute prey items (annelids), since shorebirds may have more difficulty isolathg and

capturing them. Quamrnen (1 982) found that sand interfered with shorebird capture of

very srnall prey (similar in size to the small annelids in this study). Duniin fiom WI and

MB did not differ in home range or core area sizes, nor did the densities of crustaceans or

molluscs within them differ.

Wmock and Takekawa (1996) found that the home ranges of Western

Sandpipers wintering in San Fransciso Bay were smailest at the site where feeding and

rwsting areas were in closest proximity to one another. Roost sites in the Fraser Delta

were largely ephemeral, and Dunlin fiom al1 three banding sites roosted near feeding

areas within their ranges. Thus, distance traveled to roost sites does not account for the

larger range sizes of BB birds. BB Dunlin regularly crossed the Point Roberts peninsula

between Boundary Bay and Roberts Bank, and could have thereby enlarged their home

ranges by incorporating more terrestriai 'fly-over' habitat. However, on average 43.2%

of the within-bird night detections of BB Dunlin occurred in the terrestriai habitat, and

there was some indication this average was statistically higher than those of the other two

sites (P = 0.06, Table 2.4). More importantly, the marine component of the BB Dunlin

home ranges was also larger than at the other two sites (Tables 2.1 and 2.2). Evidently,

Dunlin regulated space to use in relation to invertebrate prey density, travelling M e r

between food patches and using more space where prey density was lower.

Femaie Dunlin, the larger sex (Chapters 5 and 6), had larger marine home ranges

îhan males (Tables 2.1 and 2.2). Home range size generally increases with body size

among vertebrate species (McNab 1963, Schoener 1968, Harestad and Bmell 1979,

Peery 2000), and sex-related differences in body size can largely explain differences in

range size (Harestad and Bunnell 1979). However, within my radio-marked sample of

Dunlin, overall home range size was not significantly related to body size (PC1) (Chapter

3, Table 3.3). It is possible that the difference between the sexes in Dunlin home range

size may have been due to différences in the density of invertebrates within their

respective home ranges, however the evidence to support this is tenuous. There was

some indication that crustaceans, which are important prey items forpcifica Dunlin,

may bave been less dense within female home ranges than those of males (P = 0.06,

Table 2.8). Crustacans were significantly less dense within femaie than within male core

areas (P = 0.04, Table 2.8). Females also had lower densities (although not significantly

so) of al1 three other invertebrate types within their home ranges. The direction of the

trend was the same for ail four invertebrate types, however, M e r research would be

required to determine whether or not the home ranges of female Duniin in fact contain

lower densities of invertebrates than those of maies.

It is uniikely that males exclude females fiom areas of potentiaily higher

crustacean density, since intra-specific aggression in C. a. pacifica is rare. Individuais in

the Fraser Delta spent less than 0.5% of their time in aggressive interactions (Chapter S),

and intra-specific aggression in a population of C. a. pacifca wintering in Caiifonia was

also rare (Warnock 1994). In addition, female Dunlin are larger than males and are

thought to be dominant over them in two ecologically similar congeners, Western

sandpipers (C. mauri) and Least sandpipers (C. minutilla) (Myers 1 98 1). In any case,

Dunlin foraging in sites of high versus low prey density did not differ in the percentage of

time they spent foraging, so there may not be a particular advantage to using areas of

potentially higher crustacean density.

Home range size did not differ between Dunlin of different age classes, nor were

there any differences, either collectively or individually, in the densities of the four

invertebrate types within adult and juvenile home ranges. Therefore it appears that

juvenile and adult Dunlin modulate their use of space in a similar fashion.

Home range data wcre collected over a period of weeks, so Dunlin could be

moving through their space daily or they couid be using different parts of it fiom day to

day (a trajectory through space over time). 1 used data on mean distance moved between

consecutive high and low tides to determine whether there might be higher movement

costs associated with larger home ranges. In comparisons among sites within sexes (since

there was a significant (P = 0.01) interaction between sex and site), the only significant

difference was that BB males moved farther than MB males (Table 2.5). In comparisons

between sexes within sites, the only significant difference was that BB males moved

farther than BB females (Table 2.5).

It is ofien assumed that energetic costs increase as the arnount of space used

increases. These costs can be substantial in territorial species, which defend their spaces

(Kodric-Brown and Brown 1978, Myers et al. 1981, Schoener 1983). In contrast, the

energetic cost differential between covering relatively more or less distance between high

and low tides may not be significant. Harestad and Bunnell(1979) state: "Actual costs of

3 7

travershg the home range are generally unknown but appea. smail. Osuji (1974)

calcuiated that such costs increase the energy requirements of a 50-kg sheep only 15%."

(p. 396). BB male Dunlin moved statistically significantly farther between consecutive

high and low tides than BB femaies and MB maies. However, since movement produces

body heat that may decrease the costs of thermoregulation, the net difference in energetic

cost between Dunlin using larger versus those using smaller home ranges, within the

range of home range sizes exhibited by Fraser River Delta Dunlin, may be small.

Patterns of space use within the home range

Core area location

Dunlin core areas contained higher densities of crustaceans and small annelids

than did their home ranges, so not only did the birds use less space at sites where prey

density was higher, but they also concentrated their use of space in areas of higher food

density across sites.

Macro-habitat choice

Most (> 70%) of the radio-marked Dunlin wintering in the Fraser Delta exhibited

some switching between marine and terrestrial macro-habitats, primarily at night, and

during high tides. Dunlin that were located in the terrestrial habitat used it primarily for

foraging (more than 60% of the time spent there) (Chapter 5). On average, Dunlin spent

between 4.0 and 2.9 hours per 24-hour day in the terrestrial macro-habitat (winter and

spring, respectively). Preliminary estirnates using radio isotope techniques are that 30%

of the Fraser Delta Duniin population's diet comes h m tercestrial habitats, and some

individuals obtain as much as 92% of their food there (L. Evans-Ogden, pers cornm.).

Seventy percent is a minimum estirnate of the percentage of the Dunlin population

that exhibited habitat-switching behaviour, since the individuals that were never located

in the terrestrial habitat at night were twice as likely to be missing fiom the study area

altogether at that the . Because the radio signals could be heard fiom approximately one

quarter the distance in the terrestriai habitat than in the marine habitat due to landscape

feanues that blocked or diminished signals, 1 may have underestimateci the magnitude of

3 8

the use of terrestrial habitats at night. However, this bias does not explain the almost

complete absence of Dunlin fiom the terrestrial habitat during the day. During daytime

hi& tides, less than 10% of the radio-marked Dunlin were missing fiom the study area,

some of which were taking part in high-tide flock-flights out of range offshore (pers. obs.,

D. Dekker and R. Swanson, pers. comm.).

The availability of marine intertidal habitat decreases with the rising tide, and

increased freshwater input during periods of rainfall may cause certain invertebrates to

burrow deeper into the sediment. During some years of heavy min, part of the population

of Dunlin wintering in Bolinas Lagoon, Cdifoniia moved more than 100 km inland to

fieshwater wetlands in the Sacramento valley (Warnock et al. 1 995). Rainfall cm

negatively affect shorebird intake of marine invertebrate prey (Goss-Custard 1970b,

Goss-Custard 1984, Pienkowski 1981), while at the same time increasing the availability

of interior seasonal wetlands and of the terrestrial invertebrates that reside there

(Gerstenberg 1979, Heitrneyer et al. 1989, Wamock 1994, Colwell and Dodd 1997). In

my study, the availability of small annelids was lower dwing rainfall than during clear

weather conditions (Appendix, Table 2.A.2).

Males were more likely to be located in the terrestrial habitat (64.4 % + 5.8 of

locations) than females (43.3 % + 7.3) (FI,, = 5.1, P = 0.03).Smaller, shorter-billed

Dunlin (predominantly males) were more likely to be Iocated in the terresuiai habitat at

night than larger, longer-billed Dunlin (predominantly females). Al1 of the known-sex

Dunlin fiom Bolinas Lagoon that made large-scale movements inland were also males

(Warnock et al. 1995). The sex difference in the use of terrestrial habitat may be due to

males having diminished access to subsurface marine invertebrates and, therefore, king

more likely than femaies to switch foraging habitats.

Other authors have noted sex-related differences in shorebud habitat use that they

atûibuted to differences in bill length and access to sub-surface prey (Smith and Evans

1973, Harrington 1 982, Puttick 198 1, Townshend 198 1, McCloskey and Thompson

2000). Townshend (1981) found that shorter-billed (male) Curlews (Numenius arquata)

were more likely to forage in terrestrial habitats (as an alternative to marine intertidai)

than fernales. In detailed observations of one male and one femaie Curlew foraging in the

39

intertidal habitat, he recordeci a luwer capture rate for the male at low temperatUres, when

invertebrate prey tend to burrow more deeply (whereas the male's capture rate equaied

that of the female at higher temperatUres). Female Curlew Sandpipers (Calidris

ferrugineu), whose average bill length was 4 mm longer than tbt of males

(appmximately equal to the size difference in the radio-rnarked Duniin in this study), had

greater foraging success and foraged faster (higher probe rate) than maies (Puttick 198 1 j.

The pattern of terrestriai habitat use exhibited by Dunlin is consistent with the

hypothesis that birds make space use decisions based on an evaluation of the changing

trade-offs between foraging benefits and predation risk (see teview Lima and Dill 1989).

See chapter 5 for a discussion of relative predation risk by macro-habitat and time of day.

1 hypothesize that the terrestriai habitat is aiways riskier to Dunlin than the marine

habitat, but that the difference in risk between the two macro-habitats demases at night.

Dunlin may benefit fiom access to terrestrial habitat as an alternative feeàing site, even

assurning higher risk in that habitat, when high tides decrease the availabitity of marine

habitat or when rainfall causes marine invertebrate prey to burrow daper into the

sediment, resulting in a decrease in food intake rate. An individual Dunlin may evaluate

the trade-off between a prospective foraging benefit and predation risk based on its own

characteristics (sex, age, sub-population, size) and interna1 state (instantaneous risk of

starvation), and on the environmentai conditions at the time (tirne of day, tide stage,

weather). For example, if it is high tide at night and the individual in question perceives

its present starvation risk to be high, it may 'decide' that the benefit obtainable by

switching to forage in terrestrial habitat outweighs the cost of subjecting itself to higher

levels of disturbance and predation risk. For the same reason, if the individual in

question is a mde and it is raining, it may select the terrestrial habitat even duting low

tide.

Movements

Mean distance moved between consecutive high and low tides and the fiequency

of smd-scde laterai movements (an index of disturbance) were both g r a t a by &y than

by night, and in marine than in terrestrial habitat. These differences in movements may

40

again due in part to differences in predation riskfdisturbance, or simply to differencm in

how Dunlin react to predation riskldisturbance, by macro-habitat and time of day. During

the day, Dunlin gathered in large flocks and spent on average 3 hours in flight, mostly

during high tide. At night, they remained in small flocks and did not make high tide

flock-flights (pers. obs.). At night, a higher proportion of Dunlin in the Wadden Sea

region evaded predators by squatting in place and staying still than during the day

(Mouritsen 1992). Dunlin may resûict their movements in the higher-risk terresniai

habitat in order to attract less attention fiom predators. However, the difference in mean

movement distance by macro-habitat may aiso have been due in part to the fact that

marine habitat is subject to tidai fluctuations.

Indices of site quality

Dunlin were not disûibuted evenly throughout the areas associated with the three

banding sites in the Fraser River Delta (Table 2.1 1). Dunlin density was lowest in the BB

area, where overall food density was lowest and Dunlin home ranges were largest. Butler

(1992), who surveyed the same areas as this study (not including WI) also recorded the

lowest densities of Dunlin in the BB area. Neither food density nor home range size

differed between WI and MB; however, Dunlin density was higher at WI than MB. Do

individual Dunlin perceive the BB site, where food and conspecific densities were lowest

among the three sites in the Delta, to be of lesser quality?

The number of faicons was highest at WI, where Dunlin density was highest. The

Dunlin at WI would therefore have been subject to more fiequent attackdday than Dunlin

at either BB or MB, assurning ail falcons had similar food requkments and consumeci

similar nwnbers of Dunlin per &y. In fact, WI Dunlin did exhibit the highest small-scale

movement fkequency of the three sites. The predator distribution data do not show the

lowest predator nurnbers at BB (where Dunlin density is lowest), however, small-scde

movement fizquency (as an index of disturbance) was lowest at BB.

While 1 cannot measure fitness, or even had the power to compare survivorships

during the study period, Dunlin fiom BB and WI did not differ in size or weight (Table

2.12), or in percent time spent foraging within each sex. These results do not suggest a

4 1

measurable difference in performance between the birds fiom these two sites. Prey

ingestion and fattening rates in shorebirds can decrease with decreasing prey density,

although this is not always the case (Goss-Custard 1977% Myers et al. 1980, Mawhinney-

Gilliland 1992, Piersma et ai. 1995, but see Goss-Custard 1 WOa, 1970c, 1981). 1 have no

data on intake rates, and body weight and size measwements were taken only when the

birds were initiaily banded, so it is still possible that Dunlin fiom BB may have had lower

intake rates. However, if BB birds were experiencing deteriorating body conditions, they

would presumably have increased their time spent foraging relative to birds from WI.

1 propose that the similarity in time spent foraging between WI and BB (ûue

during daytime and throughout the 24-hour day, Chapter 5) reflected similarities in intake

rates that may have been the result of a trade-off between prey density/familiarity and the

combined effects of interference fiom conspecifics and fkquency of disturbance by

predators. Intake rate can increase with increasing prey density due to 1) higher

encounter rates (Goss-Custard 1977b, Myers et al. 1980, Piersma et ai. 1995, Turpie

1995), and 2) greater familiarity with prey distribution as a result of using less space

(Piper and Wiley 1990). On the other hand, higher densities of conspecifics are attracted

to areas of higher prey density, resulting in higher rates of interference (Goss-Custard

1976, Ens and Goss-Custard 1984). Shorebird feeding rates have been found to decrease

with the close proximity of conspecifics, decreases that wece not due to prey depletion,

but to such factors as anti-predator behaviour of mobile prey, disturbance of searching

behaviour, and aggressive encounters (Goss-Custard l97Ob, 1976, 1977c, Selman and

Goss-Custard 1988, Boates and Smith 1989,Wamock 1994. Turpie and Hockey 1996).

Intake rates may also decrease when foraging individuals are intemipted more tiequently

by predators. Dunlin foraging at BB experienced lower prey densities and had lacger

home ranges than WI birds, but they also experienced lower densities of conspecifics and

made less kquent small-scale movements. Therefore, the similarity in time spent

foraging (given sirnilar body size and weight) between WI and BB may indicate that they

perform similarly, and that there is no overall difference in quality between them, as

perceived by the Dunlin.

Conclusions

Dunlin in the Fraser River Delta were site-faithful within a season, both regionally

(to the Delta) and locally (to sites within the Delta). Variation in prey density was highly

and significantly correlated with the observed variation in Dunlin home range size among

sites, with larger home ranges occurring in areas of lower prey density. Dunlin focussed

their use of space on areas within their home ranges where prey density was above the

range average. They may also choose if and when to forage in terrestrial habitat based on

relative levels of predation risk and prey availability. For a discussion of the

conservation implications of the findings in this chapter, please see Chapter 7.

Figure 2.1. The Fraser River Delta study area.

Figure 2.2. Study area showing banding sites (WI = Westham Island, BB = Boundary

Bay, and MB = Mud Bay) and telemetry stations.

Figure 2.3. Marine intertidal micro-habitats (as detennined by consolidating mapping

data fkom Swinbanks 1979, Fisheries and Environment Canada 1977, Dunn et al. 1 993,

and McLaren and Ren 1995), and invertebrate collection transects on Roberts' Bank, in

Boundary Bay, and in Mud Bay.

Figure 2.4. Dunlin radiolocations (N = 39 birds, each dot represents a separate location)

detected during the day throughout the Fraser River Delta.

Figure 2.5. Dunlin radiolocations (N = 39 birds, each dot represents a separate location)

detected at night throughout the Fraser River Delta.

Figure 2.6. 95% home range (darkest polygon) and 30% core area (lightest polygon) of

the Duniin fkom Westharn Island carrying radio transmitter 6.059. The dots are the

individual locations used to construct the home range.

Figure 2.7. 95% home range (darkest polygon) and 30% core area (lightest polygon) of

the Duniin from Boundary Bay carrying radio transmitter 5.508. The dots are the

individual locations used to constnict the home range.

Figure 2.8. 95% home range (darkest polygon) and 30% core area (lightest polygon) of

the Duniin fiom Mud Bay carrying radio transmitter 5.884. The dots are the individual

locations used to constnict the home range.

Table 2.1. Results of 2-way ANOVAs testing the effects of site, sex, and age categories on the size of overall, marine, and terrestrial Dunlin home ranges and core areas (HR = home range, CA = core area). Results of reduced models reported where interactions not significant (P > 0.1 ).

Effect Overall Marine Terrestrial

Site (with sex) HR: F2,?j= 14.17 P < 0.001 HR: Fz,2.i= 12.5, P < 0.001 HR: * Fz,z2 = 2.6, P = 0.08

CA: F2,?j= 5.5, P = 0.007 CA: Fz,2j = 0.5, P = 0.61 CA: FL,Z4 = 3.1, P = 0.05

Sex (with site) HR: F1,24 = 4-3, P = 0.05 HR: 4.8, P = 0.04 HR: * F2,?2 = 2.6, P = 0.08

CA: FI ,2J= 1-7, Px0.19 CA: FiV2.j = 0.7, P = 0.41 CA: F1,2j = 2.7, P = 0.12

Age (with sex) HR: F I v I = 0.4* P = 0.55 HR: F iSI 1 = 0.2, P = 0.68 HR: FIUI 1 = 0.5, P = 0.50

CA: * FiaIo= 3.9, P = 0.08 CA: * FisIo= 4.5, P = 0.05 CA: F I , l l = 0.01, P =0.90

*= interaction term, see text for details.

Table 2.3. Least squares mean overall, marine, and terrestrial core area sizes 2 SE (km') by site (taking sex into account), sex (taking site into account), and age (taking sex into account). Sites with sliared letters not significantly different from each other (Bonferroni adjusted multiple t tests).

Overall Marine Terrestrial - ~~

Westham Island (N = 6) 2.1 t 0.8 "' Boundary Bay (N = 8) 4.2 + 0.7 " Mud Bay (N = 16) 1.720.5 l3

Males (N = 13)

Females (N = 15)

Adults (N = 5)

Juveniles (N = 9)

Table 2.4. Least squares mean percentages of time (f SE) D d i n fiom each site (taking tide stage into account) spent in the terresirial habitat at night @ SE), by tide stage (taking sex hto account). There was some indication that there may have been a difference among sites (FZJ4 = 4.4, P = 0.06).

Westham Island Boundary Bay Mud Bay N=7 N=9 N=2 1

High tide 77.3 2 1 1.8 86.3 2 7.7 87.4 2 7.7

Low tide 7.3 + 10.5 43.2 2 6.4 12.6 2 5.8

Table 2.5. Least-squares mean within-bird distance moved (km) between consecutive high and low tides (2 SE) by sex (taking site into account) and site (taking sex into account). For site comparisons, those with shared letters not significantly different from each other (Bonferroni adjusted multiple t tests).

Males Females ANOVA (sex cornparisons within sites)

Westham Island (N=7) 1.9 + 0.8~' 2.2 + 0 . 4 ~ F, , 5 = ~ . ~ 2 , ~ = 0 . 9 1

Boundary Bay (N=9) 4.4 1 0 . 3 ~ 2.6 +- 0.4" F1,7=8.S, P=0.04

Mud Bay (N=20) 2.9 2 0.3" 3.2 0.3" FlmlU=4.3, P=0.09

ANOVA (site com~arisons within sexes) Fz.,3=5.7. P=0.03 F1.t5=3.3. P=O.I I

Table 2.7. Least squares mean percent of 15-minute time blocks (t SE) that Dunlin made small-scale (0.1-1 .O km) and large-scale (> 1 km) lateral movements by sile (taking sex into account), sex (taking site into account), and age (taking sex in10 account). For site comparisons, those with shared letters not significanrly different from each other (Bonferroni adjusted multiple t tests). N = number of Dunlin,

% small-scale % large-scale movements N movements N

Westhani Island Boundary Bay Mud Bay

2-way ANOVA (with sex) = 5.6, P = 0.0 1 Fz,3z = 1 .O, P = 0.43

Male Feniale

2-way ANOVA (with site) = 4.7, P = 0.04

Adult Juvenile

2-way ANOVA (with sex) F1J7 = 1.0, P = 0.22

Table 2.8. The results of tests to determine whether there were differences arnong site, sex, or age categories of Dunlin in the densities (# per m3 of marine invertebrates within their home ranges and core areas. The four invertebrate types (LA = Large Annelids; SA = Small Annelids; C = Cmstacea; M = Mollusca) were tested collectively using 2-way MANOVA (ALL), and separately using 2-way ANOVA. None of the interaction tenns were statistically signifiant at the P < 0.1 level, so al1 results reported here are those of the reduced models.

Category Home Ranges Core Areas

Sex ALL k = 0.85, FdJO = 0.9, P = 0.52 b 0 . 7 5 , FJJO= 1.6, P=0.17 (with site)

LA F1y = 0.9, P = 0.36 FIU= 2.8, P = 0.55

Age ALL k=0.64,F4,8=1-l,P=0.41 = 0.75, F J , ~ = 0.7, P = 0.63 (with sex)

LA FIJI = 0.2, P =0.98 Fl,il=3.2, P=0.10

Table 2.9. Least squares mean # invertebratedm2 + SE (LA = Large Annelids; SA = Srnall Annelids; C = Crustacea; M = Mollusca) within D d u i home ranges and core areas, by site (taking sex into account), sex (taking site into account), and age (taking sex into account). For site comparisons, those with sbared letters not significantly different h m each other (Bonferroni adjusted multiple t tests).

Home Ranges Core Areas

Westbam Island "LA: 143 2 5 A ~ ~ : 131 2 10 N=6 *SA: 10570 $772 A ~ ~ : 9142 2 4919

3323 2 182 3543 I 659 AB^: 947 + 94 A ~ : 1076i316

Boundruy Bay 'LA: 1002 5 N=8 *SA: 12568 5 695

'c: 2298 164 A ~ : 894 + 85

Mud Bay C ~ ~ : 59 + 3 N=16 'SA: 18546 + 478

3435 1 13 'M: 1195 258

Males LA: 103 + 4 N=13 SA: 1408 1 .t 545

C: 3194 128 M: 1057 I 66

Females LA: 99 + 3 N=l5 SA: 13709 2 503

C: 2844 ;t 118 M: 968 461

LA: 95 + 7 SA: 177 13 2 3605 C: 4225 5 483 M: 1097 232

LA: 79 6 SA: 20580 + 3036 C: 2815 2407 M: 822 5 195

Adulîs N=5

Juveniles N=9

LA: 5 9 5 1 SA. 18316 2 935 C: 3412 2 256 M: 1287 2 130

LA: 59 +_ 1 SA: 18674 2 695 C: 3448 2 168 M: 1145 I 97

LA: 83 2 7 SA: 27109 +_ 3014 C: 6513 2 770 M: 1133 5441

LA: 68 +_ 5 SA: 2 1656 5 2239 C: 4760 2 572 M: 1 166 f 328

Table 2.1 1. Numbers of Dunlin and their predators, as well as Dunlin density estimates (numbers per km20f mudflat), for the areas associated with each of the three banding sites in the Fraser River Delta. Count data are from Christmas Bird Counts, 1995/96 ihrough 1998/99. Predators include falcons (PEFA, MERL, PRFA, GYFA, AMKE) and owls (SEOW, BAOW, SNOW), however, falcons are likely to be more important predaiors than owls (Chapter 5).

Dunlin numbers Falcon numbers Owl numbers Dunlin densities Year WI BB MB WI BB MB WI BB MB WI BB MB

Table 2.12. Cornparisons of mean male and female body size, body weight, and percent time spent foraging between Dunlin from WI and BB. included are results of one-way ANOVAs comparing body sizes and body weights, as well as two-way ANOVAs (taking tide stage into account) comparing foraging times, between sites within each sex. Foraging tirne means are least squares means taking tide stage into account.

Body size (PC 1 ) Body weight (g) % time spent foraging

Males WI: -1.5 47.0 79.4 I 4.1

BB: -1.2 I 0.4 47.2 & 0.3 84.0 + 2.2

Females WI: 1.6 2 0.2 53.1 1 .O 77.2 2 3.5

BB: 1.4k0.1 51.0+ 1.3 78.2 + 3.3

Figure 2.1

Figure 2.2

Figure 2.3

Figure 2.4

Figure 2.5

Figure 2.6

Figure 2.7

Appendix 2.A. 1. Mean # invertebrates/m2 2 SE by transecl and micro-habitat. N = number of cores. Transects and micro-habitats shown in Figure 2.3. The % total area values here are the proportion of the total area taken from the marine intertidal micro-habitat basemap (Figure 2.3).

Molluscs Cruslaceans Small Annelids Large Annelids % Total Area

Roberts Bank (N = 26) Sandy mud Muddy saiid Sand

Boundary Bay (N = 26) Algal mathand Sand Eelgrasdsand

Mud Bay (N = 17) Sandy mud Muddy sand

Median invertebrate density : 805

% Total Area > median invertebrate density: 64.1

Appendix 2.A.2. Mean # inveriebrateslcore 2 SE under dry and wet conditions in each habitat (LA = Large Annelids; SA = SmaH Annelids; C = Cnistacea; M = Mollusca), and results of paired t-tests. N = number of cores.

DJY Wei N=15 N=15 Paired t ,4 P (one-tailed test)

Sandy mud

Muddy sand

LA: 106261 LA: 149 5 83 LA: -0.1 0.48 SA: 36646 2 1480 1 SA: 2467 1 2 9764 SA: 1.9 0.04 C: 9575 2 1756 C: 7134 2 2258 C: 0.9 0.19 M: 786 2 585 M: 1083 2 521 M: 0.2 0.44

LA: 28 2 19 LA: 28 2 19 SA: 708 2 460 SA: 212 2 106 C: 142 2 58 C: 142 2 83 M: 4572 2 1275 M: 4650 2 53 1

Appendix 2.A.3. Mean # invertebrateskore + SE by time o f day in each habitat (LA = Large Annelids; SA = Small Annelids; C =

Crustacea; M = MoIlusca), and results of paired 1-tests. N = number of cores.

D ~ Y Night N=16 N=16 Paired lis P (two-tailed test)

Sandy mud LA: 112251 LA: 1595 52 SA: 28201 + 121 63 SA: 80526 + 15468 C: 9554 + 1302 C: 6322 2 2 169 M: 7162431 M: 828 + 215

Muddy sand LA: 32 & 21 LA: 2392 111 SA: 748 + 520 SA: 2675 + 980 C: 9 6 2 4 0 C: 717 + 156 M: 4698 + 1439 M: 2834 + 405

LA: -2.2 0.05 SA: 1.0 0.33 C: -2.0 0.06 M: -0.6 0.53

CHAPTER 3

THE IMPORTANCE OF PREY AVAUABILITY LN DETERMINING THE

AMOUNT OF SPACE USED BY INDIVIDUAL NON-TERRITORIAL DUNLIN

(Cdidrir alpina puc~jiia) IN WTNTER

ABSrnCT

Utilizing radio telemeûy, GIS applications, regression analyses, and

randomization models, 1 exarnined the relationship between prey availability and the

arnount of spaca used by individual non-territorial shorebirds. The amount of space used

by 29 Dunlin wintering at three sites on the Ftaser River Delta, B.C. was quantified with

estimates of marine home range and core area size. The density and variability in density

(patchiness) of large annelids 11 I cm long), small annelids (< 1 cm long), crustaceans,

and molluscs were used as measures of prey availability. Across sites, marine home

range size decreased as prey density within the home range increased, with prey density

accounting for 63% of the variance in home range size. Within a single site, both marine

home range and core area size decreased as prey density increased, with prey density

explaining 89% of the variance in home range size and 80% of the variance in core m a

size. Mer controlling for the effects of prey density, neither Dunlin marine home range

nor core area size increased with increasing prey patchinw. Individual wintering Dunlin

appear to have modulated the arnount of space they used in relation to prey density.

Residual mean time spent foraging in marine habitats (taking body size and season into

account), which was used as an index of mean food iniake rate, did not Vary with prey

(crustacean) density . Dunlin may achieve similar mean food intake rates across a range

of prey densitieslarea sizes by baIancing the costs of foraging search andior travel time

and the costs of interference cornpetition.

MTRODUCTION

Most of the empirical research examining relatioaships between animal

distribution patterns and factors such as food availability has taken place at the population

level. However, disiribution models usually assume that the mechanism by which

patterns arise is through individuai behavioural responses (Fretwell and Lucas 1970,

Fretwell 1972). Investigations of factors affecting space use by individual birds have

largely been resûicted to those species exhibiting temtoriality, due to the logistical

difficulties of following individuals that do not defend a perimeter or retain exclusive use

of space. The advent of radio telemetry technology has made it possible to determine the

size and location of spaces used by non-territorial individuals over time, and to relate

these to factors such as the distribution of food resouces. The literature on territory size

and its determinants can be used to provide a theoretical framework within which to make

predictions about factors that may influence the amount of space used by non-territorial

individuals.

Food density and temporal variability in food density are important determinants

of the size of an individual's territory (Schoener 1968, Holmes 1970, Gass et al. 1976,

Gass 1979, Kodric-Brown and Brown 1978, Myers et al. 1979, Dunk and Cooper 1994,

Turpie 1995). Territory size can be proxirnately regulated by food density, with

individuals assessing prey density and adjusting the s ix of the area they defend as needed

to satisfy their energetic requirements (Schoener 1968, Holmes 1970, Gass et al. 1976).

Territory size can also be proximately regulated by competition, but since conspecific

density, and therefore competition, is greater in areas of higher food density, temtory size

is still ultimately regulated by food density (Myers et ai. 1979, Dunk and Cooper 1994,

Turpie 1995, Tripp and Colazo 1997).

Optimal territory size is that which maximizes the positive difference between

costs and benefits. The costs of defense (in t m s of energy and tirne) ultimately iimit

temtory size and play an important role in detennining the optimum (Kodric-Brown and

Brown 1978, MacLean and Seastedt 1979, Myers et ai. 198 1, Schoener 1983, Ydenberg

and Krebs 1987). Net benefit (resource value minus costs of use, not including defense)

to the non-territorial individual, as to the territorial one, should increase with increasing

area to the maximum arnowit of resources the individual can process, and should

eventually thereafter decline (Myers et al. 1981). In non-temtorial individuals, the

arnount of space used is not constrained by the costs of defense, so the optimal a m size

should be that at which the maximum is reached, assuming there is a meaningfîd decline

in net benefit if ma use exceeds the maximum amount of resources the individual can

process. Such an assumption may not be true energetically in the absence of defense

costs, but there may be otber benefits to using only as much space as necessary, such as

enhanced familiarity with the distribution of prey and predators. Familiarity may

improve the ability of individuals to find food and avoid predation, thereby decreasing

their likelihood of mortality (Clark et al. 1993, Wamock 1 994, Dierschke 1998).

Where resource availability is higher, the maximum amount of resowces the

individual can process should be reached at iower area values (MacLean and Seastedt

1979, Schoener 1983, Turpie 1995). The amount of space used by both territorial and

non-territorial individuals shodd therefore decrease with increasing resource value.

While optimal temtory size should decrease with increasing resowce value, net benefit

(among similar individuals) is asswned to be about the same across a range of temitory

sizes (Mares and Lacher 1987). Studies of temtorial bud species have demonstrated that

while territory size decreased with food density, the total energy content of territories

remained constant, and at levels approxirnating the individuais' energetic requirements

(Gill and Wolf 1975, Gass et al. 1976, Kodric-Brown and Brown 1978, Hixon et al.

1983). Since non-temtorial individuals do not retain exclusive use of their spaces, other

individuals will have an effect on the totai energy content of those spaces. It is therefore

unlikely that individuals could accurately assess the total energy content of a particular

space in order to harmonize the arnount of am used precisely with their energetic

requirements. Non-territorial individuals should require more space (assuming range

overlap) than temtorial individuals with similar energy requirements. Further, while

temtory size decreases with increasing conspecific density, due to rising defense costs,

the arnount of area used by non-territorial individuals should increase with increasing

conspecific density, due to higher levels of food depletion (Myers et al. 1979, Dunk and

Cooper 1994, Turpie 1995, Tripp and Colazo 1997).

Previous research has show that the densities of many non-breeding shorebirds

are significantly positively related to prey density (Goss-Custard 1970a, Goss-Custard et

ai. 1977 and 1991, Bryant 1979, Rands and Barkham 198 1, Hicklin and Smith 1984,

Colwell and Landnim 1993, Piersma et ai. 1993, Yates et al. 1993). The density of

Dunlin (Calidris alpina pacijlca), a non-territorial sandpiper wintering at three sites in the

Fraser River Delta, was lowest where ovedl marine prey density was lowest (Chapter 2).

Therefore, there is some indication that individual Dunlin rnay make decisions about how

to distribute themselves in the marine intertidal zone based on an evaluation of relative

prey density.

1 quantified the amount of space used with marine home range and core area

estimates. Home ranges are areas regularly used by individuals during a particular pend

of time, such as a non-breeding season or a migratory stopover, and core areas are areas

of concentrated use within the home range (Brown 1975, Samuel et al. 1985). Home

range size generally decreases with increasing habitat productivity among vertebrate

species (Schoener 1968, Harestad and Bunnell 1979, Peery 2000). However, this

relationship has rarely been examined among individuals within a species, except in

terms of responses to seasonal changes in productivity or of differences between sexes in

sexuallydimorphic species (but see Storch 1995 and Relyea et al. 2000). Relyea et al.

(2000) state: 'ive are not aware of any study that has examined how spatial differences in

productivity affect the size of animal home rages (within species)." (p. 147).

My study was designed to determine: 1) whether individual non-temtorial Dunlin

use less space as marine invertebrate density increases, and 2) whether estimated mean

food intake rate by Dunlin differs across a range of prey densitiedarea sizes. in other

words, 1 wish to determine whether individuals employ space use strategies with regards

to food (small ranges in high prey density areas versus large ranges in low prey density

areas), and, if so, whether the strategy used affects the individual's performance, in terms

of their rate of food acquisition. Mean food intake rate was not detennined directly.

Instead, 1 used the residual mean time spent foraging in the marine macro-habitat (taking

body size and season into account) as an index of mean food intake rate. I assumed that

Dunlin were meeting their energetic requirements and that the on-the-ground energetic

costs of ?.rave1 within the range of area sizes used by Dunlin in this study did not Vary

signif~cantly (Chapter 2).

Home range and temtory sizes among vertebrate species also generally increase

with increasing body size (McNab 1963, Schoener 1968, Harestad and Bunnell 1979,

Myers et al. 1979, Peery 2000). Dunlin are sexually size-dimorphic, with females king

larger than males, and females in the Fraser Delta had larger marine home ranges than

males (Chapter 2). Harestad and Bunnell(1979) determineci that sex-related diierences

in body size among mammals appeared largely to explain diffmnces in range size, so 1

included a measure of body size as a factor in the analyses. In addition, 1 investigated

whether the amount of space used increased with increasing prey patchiness, once prey

density and Dunlin body size had ken taken into account.

METHODS AND STATISTICAL ANALYSES

The study area, capture and marking, radio telemetry, and marine invertebrate

methodologies, as well as the statistical analyses of individual home range and core area

estimates and the randomization models were described in detail in chapter 2. 1

randomized the F-statistic results fiom the regression models described below. Al1

statistical test results were considered to be significant at P < 0.05, however, results with

P-values between 0.05 and 0.1 are reported as possibly significant biologically.

1 used multiple and simple regression analyses to determine whether the marine

home range and/or core area sizes of individual Dunlin decreased as the densities of

marine invertebrates within them increased. Since home ranges and core areas were

integrated across tidal cycles and throughout the 24-hour day, they included areas that

were not used exclusively for foraging. However, Dunlin wintering in the Fraser Delta

spent similar proportions of their time foraging by day and nigfit, and spent half or more

of their tirne foraging even at roost sites during high tide (Chapter 5). Therefore, any

analysis of the relationship between the arnount of space used and invertebrate density

should involve the entire marine home range or core area.

First, 1 tested for correlations among the densities of four invertebrate types: large

annelids, small annelids, crustaceans, and molluscs. In cases where invertebrate pairs

were more than 75% correlated (P < 0.05), one member of the pair was eliminated fiom

the muitiple regression. Large annelids were involved in al1 of the significant pairings (r

> 0.84, P < 0.01) (Table 3.2), so they were deleted fiom three of the four multiple

regression models. 1 also used simple regression analyses to determine how much of the

variation in home range andor core area size could be explained separately by each

invertebrate type. Residuals were plotted to ascertain whether the regression models

adequately represented the data and to identie outliers. One significant outlier was

identified and removed fiom the home range and core area data sets.

To ensure that any relationship between invertebrate density and marine home

range andor core area size was not simply due to differences among the three sites within

the Delta where Dunlin were trappcd (essentially a three point regression with pseudo-

replication), 1 repeated the regression analyses within-site. 1 used the Mud Bay Dunlin

(MB, Chapter 2) for the within-site analyses, since 1 had by far the largest sample of

Dunlin (16) fiom this site (compared to 8 fiom Boundary Bay and 6 fiom Westham

Island). One significant outiier was identified and removed fiom the Mud Bay core area

data set.

1 used the residuals from the multiple regressions descnbed above to detenine

whether marine home range andior core area size increased with increasing patchiness,

having taken invertebrate density and Dunlin body size into account. 1 assessed

patchiness using the variation among the nine invertebrate sarnples collected from each

marine intertidal micro-habitat (see Methods section, chapter 2). 1 created an index of

invertebrate patchiness for each micro-habitat by calculating the SD/Mean of the

invertebrate density estimates. 1 used multiple regression analyses to relate the residual

marine home range and core area sizes of individual Dunlin to the indices of invertebrate

patchiness (SDIMean) within their home ranges and core areas.

Since female Dunlin (the larger sex) in the Fraser Delta had larger marine home

ranges than males (Chapter 2),I included a body size measure in the multiple regressions

relating the amount of space used to invertebrate density. 1 also used simple regression

analyses to determine how much of the variation in home range andior core area size

could be explained by body size, overall and separately by sex. As a measure of body

size, 1 performed a principal components analysis with data on culmen length, wing

length, and weight (Rising and Somers 1989). The first principal component

eigenvectors had similar signs and magnitudes for each measure (0.56-0.59), and the

cumulative correlation maîrix eigenvdue was 0.68, so PC 1 was used as my measure of

body size.

1 was unable to directly quantifi food intake rate in the population of radio-

marked Dunlin, so 1 assumed that individuals were meeting their energetic requirements

and used within-bi percent time spent foraging in the marine macro-habitat as an index

of mean food intake rates. The radios carried by the Dunlin in my population contained

activity switches, so 1 was able to collect data on foraging activity during the same period

as the data on space use (Chapter 5). 1 caiculated the mean percentage of time spent

foraging in the marine macro-habitat during the 24-hour day by each individual in the

study population. The percent of time Dunlin spent foraging varied by sex and by season

(Chapter 5). 1 therefore used the residuals of the mean percent time spent foraging once

body size (PC 1) and season were taken into account for the analysis. 1 used simple

regression analysis to detemine whether or not residual rnean foraging time in marine

habitat varied with the density of invertebrate prey within Dunlin marine home ranges

and/or core areas.

RESULTS

In analyses using data fiom al1 3 sites, marine home range size decreased with

increasing crustacean density, which was the only significant invertebrate type in the

multiple regression mode1 (Table 3.3). When home range size was regressed separately

on each invertebraie type, crustacean density was negatively related to and accounted for

59% of the variation in home range size (Figure 3.1). There was also some indication

that home range size decreased with increasing small annelid density (3 = 0.1 1, FI2, =

3.4, P = 0.07) (Figure 3.1). Dunlin body size (PCl) was not a significant factor in the

mode1 (Table 3.3).

Within the Mud Bay site only, marine home range size decreased with increasing

crustacean density, but increased with increasing mollusc density and Dunlin body size

(multiple regression, Table 3.3). When home range size was regressed separately on each

invertebrate type at Mud Bay, cnistacean and large annelid densities were both negatively

related to, and explained considemble proportions of the variation in, marine home range

size (Figure 3.2). There was aiso some indication that home range size decreased with

increasing small annelid density (? = 0.25, F,,,, = 4.6, P = 0.05) (Figure 3.2). Large

annelids were not included in the Mud Bay multiple regression model because their

densities correlateci significantly with those of crustaceans and molluscs (Table 3.2).

However, taken aione they explain a similar proportion of the variance in home range

size. When home range size was regresseâ separately on Dunlin body size (PC 1) at Mud

Bay, body size was positively related to, and explained 33% of the variation in, home

range size (FI,,, = 7.1, P = 0.01). However, when the sexes were tested separately, no

relationship between body s i x and home range size was detected (males: FI, = 1.3, P =

0.32, femaies: F i , = 0.2, P = 0.71).

Over al1 sites, the size of Dunlin marine core areas was not related to invertebrate

density, whether the invertebrate types were considered together (Table 3.3), or

separately (Figure 3.3), although there was some indication that core area size might

decrease with increasing small annelid density (8 = 0.14, FI,, = 4.3, P = 0.06). Within

Mud Bay, marine core area size decreased with increasing crustacean density, which was

the only significant invertebraie type in the multiple regression model (Table 3.3). When

core area size was regressed separately on each invertebrate type within Mud Bay,

crustacean and large annelid densities were negatively related to, and explained

considerable proportions of the variation in, marine home range size (Figure 3.4). Once

again, large annelids were not included in the Mud Bay multiple regression model

because their densities correlated so highly with those of crustaceans (Table 3.2),

however, taken alone they again explain a similar proportion of the variance in core area

size. Dunlin body size (PCI) was not significantly retated to core area size, over ail sites

or within Mud Bay (Table 3.3).

Once the effects of prey density and Dunlin body size were taken into account,

there were no effects of invertebrate patchiness on marine home range or core area size.

This was tme over al1 sites (home range: F,, = 0.2, P = 0.93, core area: F, , ,, - = 0.1, P =

0.99) and within the Mud Bay site (home range: F,, = 2.1, P = 0.14, core area: F,,, =

1.8, P = 0.23).

Residual mean foraging time in the marine macro-habitat did not Vary with the

density of prey (crustaceans) within Duniin home ranges (F,, = 0.5, P = 0.49) or core

areas (F,,, = 0.1, P = 0.77).

DISCUSSION

Individual wintering Dunlin appear to regulate the amount of space they use in

relation to prey density. Marine invertebrate density explained 63% of the variation in

marine home range size across sites, and 89% and 80% respectively of the variation in

marine home range and core area size within the Mud Bay site (Table 3.3). In examining

the relationships between the amount of space used by Dunlin and the density of each

invertebrate type separately, al1 statistically significant relationships were negative-the

amount of space used decreased with increasing invertebrate densities. Since the

relationships were significant within the Mud Bay site, where there was less variation in

both invertebrate density and the amount of space used than there was across sites, the

overall relationship was not simply due to differences among sites (essentially a three-

point regression).

Negative relationships between prey density and area use, such as those described

above, could arise by chance, if larger home ranges included more low-prey-density

habitat by nature of being larger. However, in order for this to be a viable alternative

explanation, low-prey-density areas wodd have to predorninate across the habitat

available to and used by Dunlin. Such was was not the case in the Fraser Delta study area

(Appendix 2.1 ).

Crustacean density alone explained 59% of the variation in marine home range

size overail, showing a 1.3 km2 decrease in range size with each 100/m2 increase in

crustacean density (Figure 3.1). Cnistacean density also had significant power to explain

the variation in both marine home range size (60%) and core area size (73%) within the

Mud Bay site. Crustaceans can be an important prey item in the guts ofpacijica Duniin

(Brennan et al. 1990, B. Einer, pers. comm.). Data on crustacean density may be

sufficient to predict the amount of marine habitat utilized by individual Duniin.

However, large annelid density was just as closely negatively related to both marine

home range (64%) and core area (75%) size within Mud Bay, so Dunlin may have been

cueing in to some combination of crustacean and large annelid density at that site. 1

cannot determine the relative importance of crustaceans and large annelids with the data

available, but crustacean densities appear to be more generally related to the amount of

space used by Dunlin (across sites). Dunlin marine core area size across sites was not

related to the densities of marine invertebrates, but Dunlin core areas contained higher

densities of crustaceans and small annelids than did their home ranges (Chapter 2).

Dunlin therefore focussed the use of space within their home ranges on areas of higher

prey density.

For the most part, the amount of space used was not related to body size (PC 1) in

the population of Dunlin radiernarked for this study, the exception king marine home

ranges within the Mud Bay site, which did inmase significantly with Dunlin body size.

Harestad and Bunnell(1979) attributed sex-related differences in range size among

mammais largely to differences in body size. Since 1 used culmen length (which is

highly correlated with body size in Dunlin (P<0.001) (Engelmoer and Roselaar 1998)) to

sex the birds in this study, I cannot ascertain whether this relationship was due to the

effect of sex or to the effect of body size per se. Body size in this study was not related to

marine home range size when tested separately within each sex, although by splitting my

sample of Dunlin in two, 1 reduced power to detect the relationship. My results concur

with those of Relyea et d. (2000) who, in a rare intraspecific empirical test of the body-

size hypothesis, found that mule deer home range size was related to habitat productivity

and not to body size within sexes. The lack of relationships between body size and the

arnount of space used by Dunlin in this study could also be due in part to the fact that the

magnitude of variation in area sizes used is greater than the magnitude of variation in

body six.

invertebrate patchiness was not reIated to the amount of space Duniin used once

the effects of invertebrate density and W i n body size had been taken into account.

However, prey density in the Fraser River Delta was high, so it is not surprishg that 1

was unable to detect a significant positive relationship between prey paîchiness and

marine home range or core area size.

Dunlin in the Fraser Delta appeared to Vary the amount of space used inversely

with prey density, however the time spent foraging did not vary with the density of prey

(cnistaceans) within the home rangekore area. Assuming that birds' met their daily

needs, the range of space-use strategies used by the members of the Dunlin population

appear to be similarly efficient in terms of the individual's rate of food/energy

acquisition. For temtorial individuals, similar gross energy acquisition could be achieved

across a range of prey densitiedtemtory sizes via a trade-off between mean food intake

rate and defense costs (Turpie 1995). Birds defending smaller temtories where prey

density is higher may have higher defense costs (in terms of energy and time), but ihey

may also have higher mean food intake rates (so they can eat more and faster) than birds

defending larger temtories where prey density is lower (Table 3.1). For non-territorial

individuals such as the Dunlin in this study, 1 propose that similar gros energy

acquisition (in the absence of energetic defense costs) could be achieved across a mnge of

prey densitiedarea sizes via a ûade-off between search andfor travel costs (in terms of

time) and interference competition (Figure 3.5). Such a trade-off assumes 1) that search

andlor travel cos6 (again, in terms of time) decrease with increasing prey density, and 2)

that conspecific density, and, hence, interference, increases with increasing prey density.

Non-territorial Dunlin foraging where prey density and conspecific density are

lower should spend more time searching for prey and/or travelling arnong prey patches,

but they should also experience lower levels of interference competition (Table 3.1).

Search andlor travel costs can be expected to decrease with increasing prey density due to

higher prey encounter rates and greater familiarity with prey distribution where prey

density is higher (Goss-Custard 1977b, Myers et al. 1980, Piper and Wiley 1990, Piersma

et al. 1995, Turpie 1995). Interference fiom conspecifics in close proximity can decrease

feeding rates by causing anti-predator behaviour of mobile prey, by disturtiing searching

behaviour, and by causing increased aggressive encounters (Goss-Custard 1970b, 1976,

1977c, Selman and Goss-Custard 1988, Boates and Smith 1989, Wanioc k 1994, Turpie

and Hockey 1996).

Variation in prey density did not account for al1 of the variation in Dunlin marine

home range or core area size. There was also variation among individuals in residual

mean foraging tirne. There may be a number of other factors exerting an influence over

the amount of space used by and the residual mean foraging time of Dunlin wintering in

the Fraser Delta. Dunlin were subject to the effects of tidal action on habitat availability.

Dunlin are a flocking species, so flock membership might account for some of the

variation in the home range and core area size. During high tides, Dunlin in the Fraser

River Delta roost and forage on exposed mudflat, and in and around marshes. Roost sites

are ephemeral, perhaps due in part to the fact that large, dense flocks of Dunlin forage

there (Chapter 5) and may deplete (at least temporarily) the available invertebrates.

Marine home ranges and core areas included both low tide sites used primarily for

foraging and high tide sites used for both foraging and roosting. Although the choice of

roost sites in particular may not be entirely up to the individuai in a flocking species, it

appears that Dunlin that foraged in areas of higher prey density at low tides roosted closer

to their low tide foraging sites (since marine home ranges and core areas were integrated

through time).

Earlier studies have found that shorebird density and territory size were

significantly related to prey density (Goss-Custard 1970% Goss-Custard et al. 1977 and

1991, Bryant 1979, Myers et al. 1979, Rands and Barkham 198 1, Hicklin and Smith

1984, Colwell and Landrum 1993, Piersma et al. 1993, Yates et al. 1993, Turpie 1995,

Tnpp and Colazo 1997). However, this is the first study to show that individual non-

territorial shorebirds appear to regulate the amount of space they used in relation to food

availability. Future research should investigate 1) whether individuals adjust their marine

home range andior core area sizes through time in relation to changes in prey availability,

and, if so, whether residual mean food intake rate remains constant through time, 2) how

the actuû food intake rates of individuals Vary with the amount of space used, 3) how

interference h m conspecifics at varying bird densities affects individual food intake

rates, and 4) the precise species composition of the Dunlin's diet, as well as the relative

importance of its components, and the size of prey consumeci in relation to the size range

available.

FIGURE LEGEND

Figure 3.1. Regressions showing the relationship b e e n Duniii home range size (km2)

and the densities (# per 1x12) of each of the four invertebrate types (Large and small

annelids, crustaceans, and molluscs) across sites.

Figure 3.2. Regressions showing the relationçhip between Dunlin home range size (km2)

and the densities (# per m2) of each of the four invertebrate types (Large and small

annelids, crustaceans, and molluscs) within the Mud Bay site.

Figure 3.3. Regressions showing the relationship between Dunlin core area size (km3

and the densities (# per m2) of each of the four invertebrate types (Large and small

annelids, crustaceans, and molluscs) across sites.

Figure 3.4. Regressions showing the relationship between Dunlin core area size (km2)

and the densities (# per m2) of each of the four invertebrate types (Large and small

annelids, crustaceans, and molluscs) within the Mud Bay site.

Figure 3 S. Hypothesized tradesff b e m n search andior travel time and conspecific

interference resulting in a constant mean food intake rate across a range of prey densities.

Table 3.1 : Cornparison between lemtorial and non-iemtorial individuals in expected relative behavioural responses to areas differing in food density and conspecific density.

Type of system: Territorial Non-lem torial

Observed: Food Density Conspeci fic Density

High High

Low Low

High High

Low Low

Expected: "Patch" residence time Food intake / m2 Area use Travel costs (time) Defense costs (tirne and energy) Inter ference Mean food intake rate

higher higher smaller lower higher

d a higher

lower lower larger higher lower

lower

higher lower higher lower smaller larger lower higher

d a higher lower

equal

"Patch" in this context is not a typical patch, since food density is relatively high (see chapter 3) and the distances between "patches" are relatively short.

Table 3.2. Pearson correlation coefficients for invertebrates wiihin Dunlin home ranges and core areas, al1 sites together and within the Mud Bay site alone.

L a r ~ e Annelids Small Annelids Crustaceans Molluscs

Large annelids HR

CA

Small annelids HR

CA

Crustaceans HR

CA

Molluscs HR

CA

All Sites Mud Bay AI1 Sites Mud Bay AI1 Sites Mud Bay All Sites Mud Bay All Sites Mud Bay All Sites Mud Bay Al1 Sites Mud Bay All Sites Mud Bay

*statistically significant P < 0.05 **statistically significant P < 0.01 ***statistically signi ficant P < 0.00 1

Table 3.3, Results of multiple regressions relating home range and core area size to the densities o f four invertebrate types (LA = Large Annelids; SA = Small Annelids; C = Crustacea; M = Mollusca), al1 sites together and within the Mud Bay site alone. PCI = measure of body size (incorporating weight, wing length, and culmen length).

Regression Coefficient SE P Model statistics

Home Range All Sites Int 56.55 7.12 < 0.001 SA 0.0002 0.0004 0.64 F,,, = 1 O. 1 C -0.02 0.003 < 0,001 r = 0.63 M 0.008 0.006 0.24 P < 0,001 PCI 0.85 0.97 0.39

Mud Bay Int 28.45 SA 0.0007 C -0.0 1 M 0.010 PCl 1.55

Core Area Al1 Sites Int 2.3 LA 0.009 SA -0.00003 C -0.0001 M -0.0003 PCl 0.20

Mud Bay Int 1.8 0.3 -==O.MI 1 SA 0.00002 0.0000 1 0.17 F,.,, = 9.8 C -0.0002 0.00004 O. O003 ? = 0.80 M 0.00002 0.00008 0.83 P = 0.001

00 PCl 0.05 0.05 0.36

P

Figure 3.1

Marina home range sue (km2) 4 O Q O L O Ln O

Marina home range sizs (km2) - . * N N U U L C

m o m o < 1 1 0 < 1 i o i 1 i

Marine home range size (km2)

Ô t 2 0 1 S G 1 8

Marine home range size (km2) - . d N N W W L o u o r n o r n o

Figure 3.2

Manne home range size (km2) a O

A N O 8 B

Manne home range size (km2)

Manne home range size (km2)

LII 8 0 8 s

Manne home range size (km') - d O m 8 X 8

Figure 3.3

Manne cora area size (km2)

O - b N O l U Q Y

Marine core area size (km')

O * h l U l U r n - l

Marine com ama size (km2)

Manne core ares size (km2)

: . O - ~ o r c n m - i

Figure 3.4

Marine cors ama size ( k d )

Marine a r e area size (km2) 0 - N W

Marine core area size (km2)

Prey density

HABlTAT PREFERENCES OF DUNLIN (Calidris aipina pac1j7ca) WINTERING

IN THE FRASER RlVER DELTA, B. C.

ABSTRACT

1 examined winter habitat preferences of Dunlin in the Fraser River Delta,

approximately 10 kms south of the city of Vancouver, B. C. The Fraser Delta is the

largest estuary on Canada's pacific coast and one of few North American sites where the

wintering Dunlin population appears to be stable. 1 used radio ttlemetry to quantifi

Dunlin habitat selection at two scales (regional and local) throughout the 24-huur day and

twicedaily tidal cycles. I tested for differences in selection between sex and age classes,

and among birds captured at different sites throughout the study area. Dunlin chose

habitats non-randomly at both scales, and there were significant differences arnong sex

and site categories. Dunlin preferred marine habitats, but most individuals (>80%) wtre

located in a range of terrestrial habitats as well, particularly at night. The importance of

the terrestrial habitats to Dunlin had previously been underestirnated since they are used

far more at night than during the day. Regionally, soil-based agricultural crops ranked

above the other two terrestrial habitats, and pasture was the only terrestrial habitat that

was ranked highly and was significantly preferred at both scales. Pasture vegetation

tends to be short, and Pasture fields in the Fraser Delta are fertilized heavily and naturally

with cattle manure and likely support abundant terrestrial invertebrates. I recommend

that P d a n d managers secure a mosaic of soil-based agricultural crops, with an

emphasis on pasture. Terrestrial habitat fhgmentation should be kept to a minimum,

since Dunlin preferred large fields, likely in response to predation risk.

INTRODUCTION

Declines in many North American shorebird species have been attributed to

habitat los, particularly of coastal wetland habitat, in areas used during the non-breeding

season (Paulson 1993, Page & Gill1994, Brown et al. 2000, Momson et al. 2000 and

2001, Donaldson et ai. 2001). Of the shorebird species that nest in the arctic and sub

arctic and winter prirnarily on coastal wetlands, only one sbys within North America

throughout its He cycle. The Pacific coast subspecies of the Dunlin, Calidris alpina

pac@ca, breeds in Alaska and is a common winter resident fiom southem British

Columbia to Mexico (Warnock and GiH 1996). The sub-species as a whole is thought to

be declining, perhaps because an estimated 30-91% of its winter habitat has been lost

(Warnock and Gill 19%). However, there are sites wiîhin the Dunlin's winter range

where populations appear to be stable (Paulson 1993).

The Fraser River Delta in southwestem British Columbia is one such site.

Approximately 40,000 Dunlin winter in the Delta, and over 100,000 pas through dwing

spring and fa11 migration (Butler 1992). The Fraser River Delta is located approximatel y

10 kms south of the city of Vancouver, the third-largest and one of the fastest-growing

cities in Canada. The delta also supports the highest densities of wintering waterbirds,

shorebirds and captors in Canada (Butler & Campbell 1987), and is a key wetland

stopover site for many species of migrant birds flying between breeding habitat in

Canada, Alaska and Russia and wintering habitat in southem USA and Central and South

Amerka. Over two million shorebirds use the delta annually, primarily Westem

Sandpipers (Calidris mauri) and Dunlin (Caldris alpina pacifica) (Butler & Vermeer

1994, Butler, pers. comm.).

A thorough investigation of habitat selection should incorporate data collected at

different scales and throughout the 24-hour &y. Animals may use different critena when

making decisions at different d e s , so information h m more than one scale is needed

to tùlly understand their requirements (Moms 1987, Pedlar et al. 1997, Saab 1999). For

example, Orians and Wittenberger (1991) found that female Yellow-headed Blackbirds

prefemd to nest in marshes with higher emergence rates of odonates (the most important

food for nestlings), but within the marsh, they chose nest sites based on the density of

vegetation rather thao local odonate abundance. By the same token, Dunlin could for

example be selecting winter home ranges based on resource availability (food and roost

sites), and local foraging habitats based on trade-offs between food intake and predation

risk. Shorebirds wintering in temperate estuaries are also active both day and night

(Mouritsen 1994, Wamock and Takekawa 1 996), so habitat use estimates based on data

collected only d u ~ g the day may be biasedlincomplete (Beyer and Hader 1994).

Dunlin in the Fraser Delta area are active at night and Dunlin in the Wadden Sea used

different foraging habitats by day and night (Mouritsen 1994), so data for this study were

coliected hughou t the 24-hour day.

Habitat selection has been found to differ between age and sex categories in some

shorebird species. Age-related differences have been atûibuted to cornpetitive exclusion

by dominant adults (Groves 1978, Goss-Custard et al. 1982, van der Have et al. 1984,

Cresswell 1994) and to differences in experience (Warnock t 994, Dierschke 1998,

Caldow et al. 1999). Sex-telated differences have been attributed to differences in bill

lengths and access to sub-surface prey in tactile foragers (Smith and Evans 1973,

Townshend 198 1, McCloskey and Thompson 2000). Information on di fferences in

habitat selection between age and sex categories can assist conservation planning, since

threats may vary in type and intensity throughout an area, and may impact one sex or age

category disproporîionately, thus potentially affecthg population dynarnics.

To quantifi W i n winter habitat selection in the Fraser River Delta, 1 used radio

telemetry to study individuals of known sex, and, where possible, age that were trapped at

three different sites throughout the Delta. 1 examined habitat preferences at two scales

and ihroughout both day and night. I quantifid regional habitat selection, also termed

second-order selection (Johnson 1980), which compares the proportions of habitats

within each bird's home range ("used") to the proportions of habitats within the Fraser

Delta study area ("available"). 1 alsr, measured local habitat selection, also termed third-

order selection (Johnson 1980), which compares the proportions of locations within each

habitat ("used") to the proportions of habitats within each bird's home range

('bavailable''). Hitherto, nothing was known of either the noctumal habitat preferences of

W i n in the Fraser Delta, or of their dimal preferences. Patterns exhibited by this

stable, site-faithful (Chapter 2) population of Dunlin within the declining sub-species

may provide valuable insight into the general characteristics of a healthy and productive

temperate wintering area for M i n and other shorebirds.

METHODS

See "Study myy, "Capture and marking", and "Radio telemetry" sections in

cbapter 2.

STATISTICAL ANALYSES

See "Individual home range and core area estimates". 1 used the estimates of

home range (fixed kernel95% utilization distributions) only to assess habitat selection.

Habitat rategories

A habitat base map for the study area was constructed by digitizing polygons and

merging data h m the following digital maps: British Columbia Ministry of

Environment, Lands, and Parks 1988 TRIM rnaps (1:20,000), Canadian Wildlife Service

1989 Inventory of the Wetlands of the Fraser Lowland, and Agriculture and Agi-Food

Canada 1996 Inventory of Fraser Valley Agricultural Practices (Figure 4.1).

1 identified two marine habitat categories within the study area: 1) mudff at, the

open intertidal zone that was alternately exposed and inundated by the tide, and 2) marsh,

the vegetated zone outside the dykes. Marsh habitat was found along the shoreline

between the mudflat and terrestrial habitats, on islands in the river delta, and along

sloughs that flowed out of the terrestrial habitat. 1 inchded small areas of bog habitat

within the marsh category.

1 used digital orthophotos to further characterize the rnarsh habitat locations into

the following four sub-categories: 1) 'marsh' which was ttiickly vegetated and made up

most of the consolidated marsh habitat; 2) 'marsh edge'; 3) 'slough'; and 4) 'sparse

vegetation* where Dunlin could easily feed amongst the vegetation. These sub-categories

were consolidated for use in the habitat selection analysis, however, 1 also wished to

characterize how the marsh was utilized by wintering M i n . 1 used chi-square analyses

to determine whether marsh habitat use varied by tide stage, by time of day, by sex, by

age, or by bandhg site, and 1 discuss the use of 'matsh' and 'marsh edge' habitats in

relation to their availability.

1 took tide height and tidal amplitude into account to estimate the average amount

of mudflat available during the study period. At mean low water, the tide height was O m

and there was 152.8 km2 of exposed mudflat in the study area, and at mean high water the

tide height was 4.5 1 m and thece was O km2 of exposed mudflat in the study area.

Starting at mean low water and assurning a linear incline, approxirnately 10.2 km.! of

mudflat would have been submerged with each 0.3 m increase in tide. 1 detennined the

overall average amount of mudflat available by first adding up the products of the mean

tide heights at each time of day and tide stage (day high = 4.47 m, &y low = 2.58 m,

night high = 3.75 m, and night low = 1.35 m) multiplied by the percentages of the total

number of radio locations recorded during each of the four times of dayltide stages (&y

high = 3 1%, day low = 36%, night high = 13%, and night low = 2û%), to get an average

tide height during the study of 3.08 m. Given a 10.2 km'decrease in available mudflat

( h m a starting value of 152.8 km2) with each 0.3 m increase in tide, the total average

amount of mudflat available to the Dunlin in my study was 48.6 km2. 1 confinned the

accuracy of the tide height predictions taken fiom published tide tables (Pt. Atkinson, B.

C.) using data on actual tide heights during the time of the study collected by the Instinite

of Ocean Sciences (the mean difference was 0.1 m).

There were 26 terrestrial habitat categories in the study area, which, for the initial

analysis, 1 consolidated into the following ten: 1) 'bare' were non-vegetateâ, mud fields

with some crop remains underground; 2) 'crop residue' were bare fields with above-

ground crop remains; 3) 'pasture' were short grass fields generally grazed by cattle and

fertilized naturally; 4) 'winter cover crop' included winter wheat, fa11 rye, spring barley,

oats, annual rye grass, clover, and spnng wheat planted to provide forage for waterfowl;

5) 'grassylunknown' included grasslands, forage, weedy vegetation, uncultivated

fmland, non-agricultural land such as lawns, and pieces of unclassified habitat; 6)

'wildlands' were ta11 gras fields with a few bushes and tress; 7) 'other agriculture'

included nursery crops, f i t , bemes, and some vegetables; 8) 'turf'; 9) 'greenhouses';

and 10) 'urban' included urban areas, airports, and heavily wooded areas that were

inhospitable to shorebirds. The h t four terrestrial habitat categories are crop types

rather than "habitats", but they will be termed habitats for the purposes of this study.

The home ranges and radio locations of each Dunlin were layered over the habitat

base map. 1 calcutated the proportion of each habitat type within each home range and

the proportion of each bird's locations within each habitat type. For regional habitat

selection, the habitats within the study area that were considered available to Dunlin were

those that feH within a minimum convex polygon home range of al1 telemeîry locations

fiom ail birds considered together. In other words, the available habitats were those

within the lines connecting al1 of the outennost Dunlin locations. Habitats selected by an

individual Dunlin were those that fell within its home range. For local selection,

avaitable habitats were those wittUn the individual's home range, while selected habitats

were those within which radio locations were found.

The proportions of available terrestrial and marsh habitats differed among the

t h sites, and Dunlin showed a high level of fidelity to the areas around these sites

(Chapter 2). Therefore 1 tested to see whether Dunlin would still select habitats

regionally if I restricted the definition of what was available. Habitat use in this case was

detennined h m the mean proportion of radio locations within each habitat. Habitat

availability was determined fiom 1) the proportion of each habitat within the total study

area, and 2) the proportion of each habitat within each site-specific study area (the

minimum convex poly gon sub-area around eac h bird's banding site).

Habitat preference analyses

I used compositional analysis (Aebischer et al. 1993) to examine the habitat

preferences of Dunlin in the study area. Compositional anaiysis allows for the testing of

ciiffances in habitat use among groups (ie. banding site, sex, age), and addresses

problems of statisticai non-independence by using the individual animal as the sampling

unit and by using log-ratios of habitat proportions to determine preferences. Since the

proportions of habitats within an animal's home range or the proportions of an animai's

locations within each habitat surn to one, apparent selection of one habitat may aise h m

avoidance of another. To render my data independent, 1 divideci the proportions of each

habitat type by the proportion of mudflat habitat (the results do not depend on which

habitat is used in the denominatoc). The resultant ratios were then log-transformeci.

1 used multivariate analysis of variance (MANOVA) to determine whether Dunlin

exhibited habitat preferences (Le. whether there was a statistically significant difference

between use and availability of al1 habitat log-ratios considered simultaneously). In tests

for which statistically significant differences were found, the next steps were to rank the

habitats in order of preference, and to determine which habitats were statistically

significantly preferred over which other habitats. A matrix was consûucted for each

Dunlin to calculate, for each habitat in relation to every other habitat, the difference

between the log usage ratios and the log availability ratios. Each ceil in the matrix

calculated the log-ratio of the proportion of the habitat on the y-mis that was used to the

proportion of the habitat on the x-axis that was used, minus the log-ratio of the proportion

of the habitat on the y-axis that was available to the proportion of the habitat on the x-axis

that was available. A positive result meant that the relative usage (use in relation to

availability) of the habitat on the y-mis was greater than the relative usage of the habitat

on the x-axis, and a negative result meant the opposite. The Dunlin sample means and

standard errors were then calculated for each ce11 in the matrix (Table 4.1).

Habitats were tanked in order of preference by counting the number of positive

cases in each row of the sample mean matrix. The habitat with the most positive cases

was ranked number 1 (most preferred), and the habitat with the fewest positive cases was

ranked number 8 (least preferred) (Table 4.1). Student's t-tests were used to determine

which habitat ranks differed significantly h m each other (P < 0.05) (Table 4. l), and

these are referred to using the terms 'preferred' or 'avoided'. In addition, 1 used

MANOVA to determine whether there were any differences in habitat preferences among

groups (sex, banding site, age), and, in cases where significant differences were found,

detennined preference mnks separately by group.

To determine whether field size might also be a factor in the choice of tertestrial

habitats by Dunlin, 1 used a t-test to compare the size of fields that contained Dunlin

locations to the size of fields that did not.

Results were considered to be statistically significant at P < 0.05, however, test

results with P-values between 0.05 and 0.1 are reporteci as possibly significant

biologically. interaction terms were considered to be statistically significant at P < 0.10,

since significance tests for interaction terms have lower power than those for main effects

(Littell et al. 1991). Results of 2-way ANOVAs or MANOVAs reporteci here were those

of the reduced models in cases where the interaction term was not significant, and 1 report

least-squares means taking the other factor into account.

Since samples ofien did not meet the assurnption of normaiity, 1 used Bootsûap

and randomization models to test the robustness of the MANOVA and ANOVA results.

For tests to determine whether or not Dunlin exhibited preferences, 1 perfonned bootstrap

tests, with replacement, on the F statistic results of the MANOVAs, and estimated the

95% confidence intentais of the bootstrap distributions (1000 bootstrap samples). I report

the 95 % cofidence intentais of F statistics and their associated P-values. For

cornparisons between groups, 1 tested my hypotheses by randomizing the F statistics. For

example, to randomize the results from the MANOVA testing for sex differences in

habitat preferences, 1 randomly shuffled the sex designations of the birds in the data set

and ran MANOVAs on 1000 samples generated in this way. 1 then obtained a P-value by

comparing the F statistic obtained from the original MANOVA using the actual sex

assignments to the distribution of F values obtained from the randomly generated

sarnples.

RESULTS

Home range estimates

Seventeen of the 30 Dunlin in the home range sample were present in the study

area every day 1 tracked them, and a M e r 9 were rnissing on one day only. Overall, the

average Dunlin was absent h m 18.3% 2 2.5 SE of tracking runs. Absences may have

been due to the use of inland terrestrial habitats out of my tracking range, especially at

night when these habitats were much more likely to be used (Chapter 2). During the day,

some of the absences would aiso have been due to Dunlin flock-flying far out over the

water, likely as a tactic to avoid predation (Dekker 1998 and 1999, Hotker 2000).

Dunlin home range size differed by banding site (FU, = 14.1, P < 0.001).

Boundary Bay (BB) Dunlin had a significantly larger mean home range (52.6 lad & 4.6

SE), than either Mud bay (MB) Dunlin (23.9 km2 + 3.5 SE) or Westham Island (WI)

Dunlin (19.0 km2 5.6 SE) (Cbapter 2, Tables 2.1 to 2.3). There was some indication

that male mean home range size (28.6 km2 + 3.9 SE) may have been smaller than that of

females (35.0 km' + 3.4 SE) (FI, = 4.3, P = 0.05); however, this difference was not

statisticaily significant. Juvenile mean home range size (24.1 km' i 2.2 SE) did not differ

from bat of adults (23.6 km2 + 2.9 SE) = 0.4, P = 0.55).

Regional habitat preferences

Regional habitat use by Dunlin was non-random (A = 0.009, Fil,,,= 182,4,95% CI

of F statistic (bootstrap sample) = 174.5-7450.6, P <0.001). Five of the tercestrial

habitats (wildlands, other agriculture, greenhouses, urban, turf) were avoided in relation

to al1 of the others (Table 4.2), so 1 consolidated them under the title 'other' for the

remaining analyses. Regional habitat use was still non-random once the 'other' habitats

had been consolidated (h = 0.03, F,,, = 93.9,95% CI of F statistic (bootstrap sample) =

76.5-341.9, P <0.001). Mudflat made up 58% of the mean home range (Table 4.3), and

was preferred over al1 of the other habitats (Table 4.4). The remaining habitats each

made up less than 13% of the mean home range. Marsh, crop residue, and pasture were

ranked second, third, and fourth, and were preferred over winter cover, grassylunknown,

and 'other7 habitats.

The mean size of fields containing Dunlin radio locations (0.12 km2 5 0.8 SE,

n=64) was larger than the mean size of fields available in the study area (0.08 km2 + 0.3

SE, n=739) (f, = -5.6, P c 0.001).

Regionai habitat selection differed between sex categories (Table 4.5). Both

sexes ranked mudflat îüst and preferred it to al1 the other habitat degories, and both

sexes also had crop residue followed by pasture in the top four ranks (Table 4.4).

However, there was a higher proportion of marsh habitat in the femde (13%) than in the

male (7%) home ranges, with femaies ranking marsh second and males ranking it fourth.

Females were aiso more selective than males, preferring the habitats ranked 2 and 3 to

those ranked 5 through 7, and avoiding the 'other' habitat in relation to ail of them.

Maies used the habitats ranked 2 through 6 interchangeably, preferring them only to the

'other' habitat. There was no age effect on regional habitat preferences when tested with

sex (Â=0.27, F,,= 1.9, P=0.22).

Regional habitat selection aiso differed among the three baading sites (Table 4.5).

Mudflat made up 45% of the mean home range at WI, 55% at BB, and 63% at Ml3 (Table

4.3). Mudflat was ranked fust and preferred over al1 other habitats at BB and MB, but

was ranked second at WI, where it was less prefened than marsh, but more than any of

the terrestrial habitats (Table 4.4). Marsh was ranked first at WI (30% of the mean home

range), second at BB (7%), and sixth at MB (4%). Crop residue, which was ranked third

at WI and MB (4% and 5% of the mean home range respectively), was ranked seventh at

BB (2%). Pasture was ranked second at MB (10% of the mean home range), third at BE

(6%), and only seventh at WI (3%).

1 used an alternate test of regional habitat selection within restricted site-specific

study areas to detemine whether regionai selection may simply have arisen as a result of

differences in habitat availability among sites. Regional habitat use determined by radio

locations was non-random whether 1 used the total study area (k = 0.05, F,, = 64.4,95%

CI of F statistic (bootstrap sample) = 43.2-5006.1, P <0.001) or site-specific availability

designations (h = 0.05, F,,, = 74.0,95% CI of F statistic (bootstrap sarnple) = 52.0-

1359.8, P < 0.001).

Local habitat preferences

Local habitat use by Dunlin was non-random (A = 0.07, F,, = 45.5,95% CI of F

statistic (bootstrap sarnple) = 29.8-467.2 , P G.001). Seventy-two percent of the

locations were recorded on mudflat (Table 4.3), which was preferred over al1 of the other

habitats (Table 4.4). Marsh, which was ranked second, wirh 14% of the locations, was

preferred over al1 of the terrestrial habitats except Pasture, which was ranked third with

7% of locations. The remaining terrestrial habitats each held 3% of the locations or

fewer, and were used interchangeably.

ïhere was a significant interaction between sex and site in habitat selection (Table

4.5) due to differences arnong the groups in the selection of terrestriai habitats. There

was no consistent pattern of differences between the sexes among sites, and sex itself was

not a statistically signifiant factor either before or after removal of the interaction tenn

from the model. 1 will not go into these patterns here, except to say that they highlight

variability in the ranking of terresûial habitats. There was no effect of age on local

habitat preferences when tested with sex (k=0.29, F,.~1.7, P0.26).

Local habitat selection differed arnong the three banding sites (Table 4.5).

Mudflat made up 40% of the locations at WI, 69% at BB, and 85% at MB (Table 4.3).

Mudflat was ranked first at MB, where it was preferred over al1 other habitats (Table 4.4).

Mudflat was also ranked first at BB, but it was not preferred over marsh, which was

ranked second and made up 9% of locations. Mudflat and marsh together were preferred

over al1 of the terrestrial habitats (Table 4.4). Mudflat was ranked second at Wi, where it

was less prefemd than marsh (5 1% of locations) but more preferred than any of the

terrestrial habitats except bue, which ranked third. Bare was the only terrestrial habitat

that was preferred in relation to the others at W. At BB, 'other' was ranked last and the

remaining terrestrial habitats were used interchangeably. Pasture was ranked second at

MB (9% of locations) and did not differ fiom marsh (which was ranked third with 3% of

locations), but was preferred over grassy/wiknown, crop residue, winter cover, and bare

habitats.

Use of the mawh habitat

Marsh core was more available than marsh edge, however half of the marsh

locations were in the marsh edge category, and only 30% were in the marsh COR. Sparse

vegetation was as available as marsh edge, however only 14% of the locations were in

sparse vegeiation. The remaining 6% of locations were in sloughs, the least available

marsh habitat. Use of the marsh categories differed by time of &y (X2= 15.8, d f= 3, P =

0.001) and by tide (XZ= 19.4, df = 3, P = 0.001). Marsh core and marsh edge were used

less at night and at low tide, and sparse vegetation was used more at low tide. During the

day, marsh core and marsh edge were used significantly more than sparse vegetation and

slough (XZ = 59.2, df = 3, P = 0.001), but there was no difference in use of the four marsh

categories at night (XL= 0.5, df = 3, P = 0.91). There were no sex (X2= 4.9, df = 3, P =

0.18), age (X2 = 0.5, df = 1, P = 0.82), or site (x2= 6.9, df = 6, P = 0.33) differences in use

of the marsh habitats.

DISCUSSION

Dunlin in the Fraser River Delta selected habitats non-randornly at both regional

and local scales. They showed a significant preference for marine habitats throughout the

study area, but also used a range of habitats within the terrestrial zone (Table 4.3).

Duniin showed a high degree of within-winter site fidelity, and birds at al1 three sites

made use of terrestrial as well as marine habitats.

Overall habitat preferences

Mudflat ranked first and was preferred over al1 other habitats at both regional and

local scales (Table 4.4). The mudflat habitat in the Fraser River Delta supports high

densities of invertebrates during the winter (Chapter 2, McEwm and Farr 1986, Baldwin

and Loworn 1994), and provides the best vantage for observing the approach of potential

predators (Chapter 5). It is perhaps these factors that account for it king the most

important habitat for wintering Dunlin.

Marsh was ranked second at both regional and local scales, although it was not

preferred over pasture or crop residue. Marshes provide a refuge for Dunlin roosting at

high tide. Within the marsh habitat, marsh edge was the most selected marsh sub-

category, even though it was less available than marsh core. During the day, Duniin must

be able to take off quickly in order to evade predators. Peregrines hunting Dunlin in the

Fraser Delta were successful on 33% of 15 surprise attacks and only 8% of 287 aerial

chases (Dekker 1998 and 1999). Roosting dong the edges of the marsh rather than in the

core likely facilitates take-off in the event of a disturbance. At night, Dunlin were more

dispersed and did not gather into large flocks or use group aerial manoevres to evade

predation @ers. obs., L. Evans-Ogden, pers. comrn.), and during that tirne, both marsh

edge and rnarsh core were used significantly less.

Under regional selection, ihe soil-based agicultural crops (bare, crop residue,

pasture, and winter cover) were preferred to the other two terrestrial habitats

(grassy/unknown and 'other', which included greenhouses), both of which were less

preferred than al1 but winter cover. Pasture was the only terrestrial habitat that was

ranked high and preferred at both d e s . Pastures in the Fraser Delta are fertilized

heavily and naturally with cade manure, and they likely support higher densities of

terrestrial invertebrates compared to crop fields (Fratello et al. 1989). Pasture vegeîation

tends to be short, and densities of Dunlin using pastures in Califomia correlated

negatively wiih vegetation height (Colwell and Dodd 1995). Bare fields may attract

Dunlin because they look just like mudflats, as do some of the crop residue fields (but

with a little additional vegeîation). Winter cover crops may be less attractive to Dunlin

because they can grow quite taII, obscuring the view of approaching predators, however,

they may be used once the waterfowl wintering in the area have grazed back the

vegetation (Taitt 1997).

Regional and local preference orders differed by three ranks for al1 terrestrial

habitats except pasture and winter cover. Therefore it appears that Dunlin selected home

ranges composed primarily of mudflat and marsh, but they also preferred soil-based

agricultural land, and pasture in particuiar. Locally, Dunlin again ranked the marine

habitats at the top. They also stiowed a preference for pasture, but were not particularly

selective of the other terrestrial habitats within their home ranges.

Some of the Dunlin were not located in terrestrial habitats during this study (<

20% of the radio-marked individuals for whom I had a sufficient sample of locations to

calculate a home range). However, these birds had at least some terrestriai habitat

incorporated into their home ranges, since the home range represents the area within

which an individual Dunlin has a 95% probability of being located, and the two macro-

habitats were adjacent. One of the advantages of compositional analysis is that it uses the

individual as the sampling unit and incorporates the variation arnong individuals in

detennining mean habitat preferences.

Sex-speciîic habitat preferencea

Males and female Dunlin differed in their regional habitat preferences. Females

were more selective in their choice of habitats, exhibiting more statistically significant

preferences between habitat ranks. Females were also more efficient foragers than males

(Chapter S), and may have had to be more selective in their choice of habitats to

accomplish this. Females had more marsh habitat in their home ranges, perhaps due to

the fact that there was some indication they were more likely to be located in marine than

in terrestrial habitats (Chapter 2).

Site-specific habitat preferences

Dunlin fiom different sites within the study area selected different habitats at both

regional and local scales. Regionally, site-specific differences in habitat selection cannot

be attributed to differences in habitat availability among sites, since the results were the

same whether 1 considered the study area as a whole or just the areas around each site to

be available habitat.

BB and MB Dunlin ranked mudflat first at both scaies, but WI birds ranked it

second to marsh at both scales (Table 4.4). Regionally, the fact that WI birds ranked

marsh first over mudflat can be explained in part by the fact that there was much more

marsh habitat available at M than at the other two sites. This is because the proportion

of marsh habitat available regionally is calculated over the whole study area, including

the parts of the study area where there is little marsh (Table 4.3 and Figure 4.1).

However, since marsh was also ranked and preferred over mudflat loçally, selection of

marsh at WI is not simply due to higher availability. The tidai regime in the Fraser River

Delta is such that the mudflat habitat on Roberts' bank (the WI ma) is submerged for

approximately 1 to 2 hours longer than the mudflat habitat in Boundary and Mud bays,

perhaps explainhg the greater use of marsh habitat by the Wi birds.

Marsh was ranked the same at both regionai and local scaies within both WI and

BB Dunlin (fht at WI and second at BB), but MB birds ranked it sixth regionally and

third locally. The low regional ranking at MB is likely due to the fact that there was less

marsh habitat available for incorporation into home ranges at Mi3 than either BB or W.

The jump up to third in the local rank reflects the generai importance of marsh habitat to

Dunlin regadess of its availability.

There was considetable variation in the site-specific ranking of terrestrial habitats

both among sites and between scales, and many of the habitats that ranked high were not

statistically significantly preferred over those ranked low. 1 conclude that W i n are

flexible in their use of the terrestrial habitats, and that maintaining a mosaic of soil-based

agricuitural crops throughout the study area should meet their needs.

Conservation Implications

Mudflat was the most important habitat for wintering Duniin in the Fraser River

Delta, followed by marsh and Pasture. The importance of the marine intertidal habitats

was not in doubt, however, the importance of ihe terrestrial habitats had previously been

underestimated until they were discovered to be used far more at night than during the

day. Dunlin that were iocated in the terresaial zone used that habitat primarîly during

high tide when less intertidal habitat was available, however, the terrestrial zone was not

simply used as toosting habitat (Chapters 2 and 5). Dunlin in the terrestrial zone spent

most of their tirne foraging (more than 60% of the rime spent ihere) (Chapter 5). Early

indications h m L. Evans-Ogden's research estimates chat 30% of the Fraser Delta

Dunlin population's diet cornes fiom terrestrial habitats, and some individuais obtain as

much as 92% of their food there (L. Evans-Ogden, pers comrn.).

For some of the Dunlin wintering in the Fraser Delta, where energetic costs are

high, access to nearby terrestrial habitats may be required for hem to meet their energy

needs (Davidson and Evans 1986). Several oîher studies have asserted the importance of

alternative high tide foraging habitats, in particular soil-based agr icu i td fields, for

wintering shorebirds (Velasquez and Hockey 1991, Davidson and Evans 1986, Colwell

and Dodd 1995, Rottenburn 1946, Weber and Haig 1996, Butler 1999, Dann 1999).

Loworn and Baldwin (19%) found that intertidal habitais with adjacent farmiand

supported between 75 and 94% of four waterfowl species wiatering in the Puget Sound

region, and that few sites without adjacent farmland supportecl significant populations.

1 05

Predicted mortality rates of Oystercatchers (Haematopus ostralegus) wintering in a

similar environment in Britain increased significantly when upshore and field foraging

areas were removed h m the mode1 (Stillman et al. 2000).

Within the terrestrial zone, Dunlin preferred pasture, but they also selected a

mosaic of soil-based agricultural crops. The ranking of terrestrial habitats varied between

regional and local scales, and among birds from different sites within the study area

Such variation rnay be due in part to the fact that the sample of Dunlin used in my study

was fairly small(30), and even smaller when broken down by site (6,8, and 16). There

were eight habitat categories, some of which were available in small proportions (Table

4.3), and ofien the ranks could not be separated statistically. Altematively, the habitat

mosaic itself rnay be what Dunlin are selecting. Each terrestrial habitat rnay contain the

resources to satisfy different nutritional requirements, and rnay aiso offer different trade-

offs between energy intake, energy expenditure, and predation risk. Dunlin rnay select

terrestrial habitats based on their interna1 state at a given moment, and securing a range of

available habitats rnay be the best strategy. It is possible that Dunlin in the Fraser River

Delta are actually selecting landscape diversity, and that the population has remaineâ

stable because of the availability of a mosaic of soil-based agricultural crops adjacent to

the intertidal zone.

Agriculture is often detrimental to endemic flora and fauna, but many coastal

areas have already been altered for agricultural use. Maintainhg diversity in such agro-

ecosystems rnay therefore be the key to maintainhg diverse and stable populations of

wintering birds, as well as other taxa (Vandermeer and Perfecto 1997). Agro-ecosystem

diversity rnay also be important for the long-term stability of single species such as the

Dunlin, particularly now, when most changes to existing agro-ecosystems consist of

industrial and housing development. In the case of the Fraser Delta, there has been a

proliferation of industriai greenhouse operations (from 101 ha (1.4 % of the mil-based

agricultural land) to 154 ha (2.1%) in less than two years (data h m Corporation of Delta

building permit applications)), which, aithough they meet the land use requirements of

the Agricultural Land Reserve, essentiaily remove habitat that had been available to

wildlife and fragment the habitat that cemains.

The selection of terrestrial habitats by Dunlin in this study may have been

influenced by factors other than ground cover. The mean size of fields used by Dunlin

was 50% larger than the mean size of available fields, perhaps because targer fields

provide more open space and therefore allow for greater aàvance waming of an

approaching predator. The selection of terrestrial habitats may also have been influenced

by factors such as standing water levels and proximity to roads, as well as by the history

of land use in eacb habitat block. For example, a base field that had previously k e n

fallow and had recently been plowed under might be more attractive han one that had

been intensely cultivated in recent years. Change in vegetation height resulting h m

waterfowl grazing is another factor that may have influenced terrestrial habitat selection

(Taitt 1997); however, the Dunlin in this study were tracked over short tirne periods (a

matter of weeks), so 1 am assuming that the landscape would not have changed a great

deal within each tracking bout. Future studies will hopefull y adàress these issues and

determine whether any of these factors significantly influenced the selection of habitais

by Dunlin.

In the meantime, for Dunlin in the Fraser River Delta and other wintering areas

near to agriculturai lands, I recommend that managers secure a mosaic of soil-based

agriculturai crops, with an emphasis on naturally-fertilized Pasture, on land near intertidal

mudflats. Field fragmentation should be kept to a minimum, since Dunlin preferred large

fields, likely in response to predation risk. To-date, avian management plans for the

terrestrial habitats in the Fraser River Delta have focused on the needs of raptors,

waierfowl, and passerines, birds whose requirements differ k m those of Dunlin, so

management plans for the am will require modification. Further research will determine

1) how much the terrestrial habitat contributes to the Duniin population's energy budget,

2) whether the composition of the habitat mosaic must adhere to a particdar pattern, and

3) whether a more intricate plan is required for the long-tenn health of Dunlin and other

avian populations wintering in the Fraser River Delta.

FIGURF, LEGEND

Figure 4.1. Habitat base map showing categories usai in the habitat preference analyses.

Table 4.1. Ranking matrix for overall regional habitat preferences (al1 30 Dunlin). Values are the mean differences SE) between the log usage ratios and the log availability ratios for each habitat in relation to every other habitat. Habitats were ranked in order o f preference by counting the number of positive cases in each row of the niatrix. Positive cases mean that the relative usage (use in relation to availability) of the habitat on the y-axis is yreater than the relative usage of the habitats on the x-axis. The habitat with the most positive cases was ranked number 1 (most preferred) and the habitat with the least positive cases was ranked nurnber 8 (least preferred). Ranks that share a superscript are not statistically different from each other (P > 0.05).

Crop Grassyl Winter Mudflat (1) Marsh (2) Bare (3) Residue (4) Unknown (5) Pasture (6) Cover (7) Other (8) Rank

1.28 2 O. IO* 1 .O0 + 0.07* 0.34+0.15* 0.05+-0.17

-0.2920.1 l* 0.292 0.11* -0.35 2 O. 15* -0.63 + O. 18* 0.2220.16 -0.0720.12 -0.30 + 0.16 -0.58 2 0.12* - 1.93 2 0.29* -2.2 1 2 0.36*

*Student's t-test results showing significant differences between habitat pairs (P < 0.05).

Table 4.2. Proporlions of unconsolidated habitats available in the study area, mean proportions of unconsolidated habitats within Dunlin home ranges, and rank order o f unconsoiidated habitats within Dunlin home ranges (regional selection). Ranks that share a superscript are not statisiically different from each other (P > 0.05). Ranks numbered 1 (most preferred) through 7 are selected, relative to rariks 8 through 1 2 which are avoided.

Habitat type Available Home range Rank

Mudflat Marsh Crop residue Pasture Bare Winter cover Grassyhnknown Wildlands Other Agriculture Greenhouses Urban Turf

Table 4.4. Rank order of habitats within Dunlin home ranges, and detennined from Dunlin radio locations. Rank # 1 is the most preferred, and ranks that share a superscript are not statislically different from each other (P > 0.05). Overall and site-specific ranks are reporied, with sample sizes in brackets.

Crop Grassy/ Winter Mudflat Marsh Bare Residue Unknown Pasture Cover Other

Available: 1 5 6 7 2 4 8 3

Home Ranges: 0vera11(30) 1 A 2" 5 r ~ Y3 7~ 4DC 6t>" 8 1:

Male (1 3) 1 A 4"C 9' 6°C 3!3‘- 2''

5c'l> fD 8"

Female(l5) 1 A 2" 3" 7" qBC gCD 8"

Westham Island (6) 2' 1 A 5" 3" 6 " 7" 4C 8"

Boundary Bay (8) 1 A 313" 4BC' 7" 5"" 3U' - gC gBC

Mud Bay ( 1 6) 1 A 6" 5 3" 7" 2B 4D 8~

Radio Locations: OveraU (30) 1" 2" 8'' 6c1j 4CD 3 RC' 7D 5 c ~

Westham Island (6) 2" 1 A 3""

?Al3 9 3 " 6" ,IK'

7(. 5' 8' 4C 6!3' Boundary Bay (8) 1 A -

3bc' 8 1 6t'1>1. 5c"> qBC 7DE

8' Mud Bay (16) 1 A 2" 4 ~ c -

Table 4.5. Results of MANOVA randomizations testing for regional and local habitai seleciion in Dunlin among sex and site categories. Regional selection was determined from the proportions of habitais within the home range, and availability from the proportions of habitats within the entire study area. Local seleciion was determined fiom the mean proportion of radio locations within each habitat, and availability from the proportion of each habitat within each home range.

P-value P-value Scale Groui, A F d f Full Model Reduced Mode1

Regional: Sex 0.40 3.4 7,16 0.1 1 0.01 Site 0.03 11.4 14,32 <0.001 <O.OO 1 Interaction 0.3 1 1.8 14,32 O. 16 --

Local: Sex 0.8 1 0.5 7.16 0.80 0.5 1 Site 0.19 3 .O 14,32 0.006 0.002 Interaction 0.30 1.9 14,32 0.08 --

Figure 4.1

CHAPTER 5

TIME-ACTIVITY PATTERNS AM) THE IMPORTANCE OF NOCTURNAL

FORAGING TO HIGH-LATITUDE TEMPERATE WINTERING DUNLIN

(Calidris alpina puc~jica)

ABSTRACT

1 used radio telemetry to detennine how much time individual Dunlin (Calidris

alpina pacifica) of known sex, and, where possible, age, spent foraging throughout the

24-hour day in the Fraser River Delta, B. C., the most northerly site supporting a

significant population in winter. I also used daytime observationai sarnpling to detennine

how much time Dunlin spent on other activities such as roosting, preening, vigilance, and

aggression. Dunlin spent on average at least 15.7 hows per 24-hour day foraging

(depending on season), and at Ieast another 3 hours per day flying. This left at most 5.3

hours per day for other activities. The percentage of time that Dunlin spent foraging did

not differ between day and night, nor between marine and terrestriai rnacro-habitats.

Dunlin spent on average at least 7.1 hours foraging at night, and at least 2.9 night hours

foraging in the terrestrial macro-habitat. Dunlin speni more tirne foraging in 1996 than in

1998, and in spring than in winter, both during the day and at night. There was no

foraging time difference between age classes. Females were larger than males, but they

spent less time foraging. Assuming that both sexes were either meeting their energy

requirements, or losing and gaining weight at the same rate (Kaiser and Gillingharn

1985), the fact that females spent less time foraging indicates that they either consumed

more, bigger, or better quaiity food per unit time; females did not conserve energy by

spending less time in flight.

INTRODUCTION

Non-breeding birds need food energy for three main categocies of activity: self-

maintenance (including thennoregulation), flight (to and 601x1 foraging habitats and in

avoidance of predators), and foraging (Evans 1976). In spring, energy may also be

required for molt and the accumulation of fat in preparation for migration. The amount of

energy required, and the the-docation strategies used, can be affected by both interna1

factors, such as sex, age, and body size, and external factors, such as season, t h e of &y,

food availability, and predation risk (Evans 1976, Puttick 1984). Cornparisons of time-

activity budgets arnong diffant groups of birds, or among birds subject to different

Il6

extemal factors can "elucidate the adaptive significance of the species' activity pattern"

since b'evolution preswnably has produced optimal strategies to allocate time to different

energy-demanding and energysbtaining activities, strategies that should be expressed in

the observed time-activity pattern" (p. 783, Ashkenazie and Safriel 1979).

Behavioural data, such as time spent collecting food and avoiding predators, can

also assist in determining the relative importance of selected habitats to animal

populations. In species that are active at night, these data must be collected throughout

the 24-hour day to provide complete information on the species' requiremenis. Some

species use different habitats by day and night (Chapter 2), and noctumal activity may

contribute significantly to overall activity and energy budgets. However, tirne-activity

data are ohen lacking h m studies of habitat preferences, and are rarely collected at

night, due to obvious logistical difficulties. In this paper 1 examine 24-hour time-activity

budgets of individual Dunlin (Calidris alpina pacifrca) in the Fraser River Delta, B. C.,

the most northerly site supporting a significant population in winter.

Noctumal foraging has frequently been reported in non-breeding shorebirds, both

temperate and tropical (see review in McNeil et al. 1992). Traditionally, the role of

nocturnal foraging has been considered in light of two main hypotheses. The

'supplementary' hypothesis states that birds feed at night only when they are unable to

meet their energetic requirements during the day, and the 'preference' hypothesis states

that birds prefer to feed at night because feeding is more profitable or safer at that time

(McNeil et al. 1992).

At temperate latitudes, nochunai foraging has k e n interpreted as a response to

decreased lemperature (causing increased energy demands and decreased food

availability), and shorter daylength in winter. Together, these factors are thought to make

it dificdt for shorebirds, which are also subject to the effect of tide cycles on habitat

availability (Burger 1984), to meet their energetic requirements during the day (Goss-

Custard 1969, Heppleston 197 1, Dugan 198 1, Pienkowski 1982).

Alternatively, nochunal foraging may actually be preferred, in response to

increased prey availability or decreased risk of predation at night (McNeil et al. 1992).

Several invertebrate species, both temperate and tropical, move up the sediment column

or are more active at ni& increasing their availability to shorebirds (Dugan 198 1,

Townsend et al. 1984, Evans 1987, Robert and McNeil1989). Some Grey Plovers

(Pluvialis squatarola) in Britain met the majority of their energetic requirements at night,

due to the exclusively noctumal emergence of preferred polychaete prey Ougan 198 1,

Townsend et al. 1984). Wilson's Plovers (Charadrius wilsonia cinnamominus) wintering

in Venezuela switched to foraging primarily at night apparently due to an increase in the

risk of predation during the months that their diurnai predators (Peregrine Falcons (Falco

peregrinus)) were present (Thibault and McNeil 1994).

To provide equal emphasis to both nocturnal and diurnal foraging decisions,

Sitters (2000) restated the 'supplementary' and 'preference' as the 'choice for day'

(maximizing time foraging during the day and minimizing time foraging at night) and

'choice for night' (maximizing time foraging at night and minimizing time foraging

during the &y) hypotheses. Sitters (2000) proposed three additional hypotheses to

explain patterns of diuniai and nocturnal foraging in intertidally-feeding shorebirds: birds

may be indifferent to the time of &y and forage for the same proportion of time by day

and night (the 'indifference' hypothesis), birds rnay spread their foraging time evenly

between day and night to avoid peaks of body weighi that may make them more

vulnerable to avian predators ('feed day and night to avoid weight-related risk

hypothesis), or birds may simply forage day and night whenever not prevented fiom

doing so ('feed whenever possible' hypothesis) (see Sitters 2000 for detailed

explanations).

Foraging activity, and the relative contribution of diumal and nocturnal food

intake, c m Vary seasonally, due to variation in the environmental conditions or the

energetic requirements of the bird (iei pre-, ps t - or during migration) (Goss-Custard

1969, Momer and McNeil1991, Dodd and Colwell1996, Rompre and McNeil1994).

The relative importance of nocturnal foraging can also vary among species (Robert et ai.

1989, Zwarts et al. 1990, Dodd and Colwell 1998). However, there have been few direct

estimates of the extent to which individual non-breeding shorebirds forage throughout the

night (but see Turpie and Hockey 1996, Sitters 2000), and, to my knowledge, none using

smail shorebird species (around 50 g or less). In addition, almost nothing is known about

variation in the extent of night foraging among individuals with respect to age (but see

Goss-Custard et al. 1987) or to sex.

Females of many shorebird species are larger than males. Among vertebrate

species of sirnilar trophic levels, and specifically among shorebird species, larger species

tend to spend less time feeding than smaller ones (Gibb 1954, Calder 1974, King 1974,

Goudie and Piatt 1990, Pienkowski 198 1, Eng lemoer et al. 1984, Zwarts et al. 1990).

This has been explained as follows: among species, gut size and energy consumed per

unit time scale isometricdly with average body mas, while metabolic rate scales to the

power of approximately 0.75 (Calder 1974, Zwarts et al. 1990). Since energy intake rate

should be proportional to metabolic rate for maintenance of body mass, time spent

feeding should scale to the power of 4 .25 . In other words, large and small species will

take the same amount of time to fiIl their guts, by feeding on food of different sizes

(Willson 1971). However, since gut size scales isometrically, and since smaller species

have higher mass-specific metabolic costs, the smaller species will use up the energy it

consumed and recommence foraging before the larger species.

Among 14 shorebird species wintering in Mauritania, including Dunlin, energy

intake rate scaled very close to unity (0.95), and time spent feeding scaled to the power of

-0.22 (Zwarts et al. 1990). This relationship is rarely examined within species, since

body size and foraging niche differences between sexes may not be suficient to

demonstrate it. Female Capercaillie, (Tetrao urogalh) which are about haif the size of

males, spent significantly more time (26%) f d i n g than males in mid-winter (Gjerde and

Wegge 1987). This was attributed to the higher relative heat loss of hens, assurning

males ingested twice as much food energy per unit tirne.

The metabolic scaling factor can Vary a great deal within species, from MO5' for

Prairie Falcon (Falco mexicanus) (Kaiser and Bucher 1985) to MI.'' for Redshank

(Tringa rotanus) (Scott et al. 1996). As seen in Redshank, both BMR and

thennoregdatory costs scaled isometrically in Dunlin wintering at temperatures between

O and 34°C in California (J. Kelly, unpubl. data). If the metabolic costs within a species

scale isometrically, and both sexes forage h m the same size spectrum of prey, then the

tirne spent feeding might also scale to unity. C. a. pacifca feed on smail food items

1

(E3uchannan et al. 1983, Brennan et al. 1990, B. Elner, pers. com.) that should be easily

handled by either sex. In this paper, 1 will test the hypothesis that female Dunlin may

have to spend more tirne feedig than males to meet their higher energetic requirements,

and, if al1 daylight hours are spent foraging in winter, females may also have to spend

relatively more time foraging at night than males.

Juveniles may be less efficient foragers than adults (Groves 1978, Puttick 1979,

Goss-Custard and Le V. Dit Dure11 1987% Marchetti and Pnce 1989, Hockey et al. 1998,

Caldow et al. 1999). 1 will therefore also test the hypothesis that if al1 daylight hours are

spent foraging in winter, and if juveniles are less efficient foragers than adults (a

difference that might be exacerbated at night when visual cues to prey are decreased or

absent), juveniles may have to spend relatively more time foraging at night than adults

(Puttick 1979).

In this study 1 examine 24-hour time-activity budgets of individual Dunlin of

known sex, and, where possible, age. 1 test for activity differences by time of day, by

macro-habitat, by season, by year, and between sex and age classes of Dunlin. In

addition, 1 use data on diunial and nocturnal foraging patterns in Dunlin to evaluate the

five hypotheses outlined by Sitters (2000) for the role of night-foraging in intertidal

shorebirds. Current radio telemetry technology makes it possible to follow individuals

and record basic activity data through time (Exo 1993, Exo et al. 1992). 1 am able not

only to determine habitat preferences, but also to determine whether or not said habitats

are used for foraging, and 1 can do so throughout both day and night. By following

individuals and calculating within-bird means by tide stage, macm-habitat, and time of

day, I can eliminate biases in sampling and get a complete picture of the individuals'

behavioural strategies.

METHODS

Study area

See chapter 2.

Capture and marking

Seventy-two Dunlin were captured and fitîed with radio transmitîers to assess

activity during t h m time periods: 24 in winter 1995/%,23 in spring 1996, and 25 in

spring 1998. The same methodology us& to capture and mark the first two groups of

birds (described in chapter 2) was used to trap, measure, and mark the third group at

Brunswick Point in 1998 (Chapter 2, Figure 2.1). Transmitters were placed on 7 males, 6

females, and one bird of unknown sex at the end of January, and 9 males and 2 females at

the end of February 1998.

Radio telemetry

1 used radio telemetry and direct observation to determine the amount of time

individual Dunlin spent foraging and flying. Each radio transmitter contained a mercury

activity switch that doubled the pulse rate when the top of the radio tipped below

horizontal (in relation to the ground) (see Exo 1993, Exo et al. 1992).

1 performed a preliminary study using two captive Dunlin (part of a separate

project), to determine the best location on the birds' back for radio attachent, and to test

the accuracy of the activity switch in signaling foraging behaviour. The best radio-

attachent location to ensure that an elevated puise rate would indicate feeding activity

was one centimeter above the uropygial gland on the midline of the bird's back. 1

recorded pulse rates while observing the behaviour of one of the captive Dunlin and one

radio-marked Dunlin located visually in the field. Foraging Dunlin raise and lower their

heads and toms, taking steps and searching berneen pecks and probes. This causes

elevation of the pulse rate within the data collection interval. The activity switches were

100% accurate in signaling when the two activityeitch test birds were feeding.

Prior to attaching radio transmitters to Dunlin, the pulse rates (number of puises

per 16-second interval) of each radio at each of the two switch positions were recorded

three to five times to detennine baseline pulse rates. Among radios, the number of pulses

per 16-second interval varied h m 4 to 8 at the non-fading position, and fiom 9 to 15 at

the feeding position. Within radios, the number of pulses recorded at each switch

position during the data recording interval varied by one pulse, due to the fact that the

interval could start during either the pulse or the silence in between pulses. Birds were

therefore classified as "foraging" if the number of pulses per 16-second interval exceeded

the minimum nurnber recorded at the non-feeding position by at least two pulses. This

may result in a slight underestimate of the actuai amount of time spent feeding, but it was

preferable to an overestimate.

After attachment, and a 3day adjustment period, activity data were collected by

recording the pulse rate and the general location of each Dunlin in a sample every 15

minutes. Precise locations were tnangulaied every two hours as described in chapter 2.

Activity data were collected both day and night, across tidal cycles, and in both marine

and terrestrial macro-habitats throughout three sampling periods; winter 1995196, spring

1996, and spring 1998. During winter and spring 1995196,I collected approximately 24

hours of activity data per week, dong with location data used to assess space use and

habitat selection (Chapters 2-4). 1 collected 67 hours of activity data h m 29 December

to 20 January l995/96 (winter), and 64 hours h m 28 Fe bruary to 20 March in 1996

(spring). In spring 1998,I added 24 telemetry stations (primarily inland) to the 61

already in existence to get better coverage of the Brunswick Point birds (Chapter 2,

Figure 2.2), and focussed on the collection of activity data. Observers collected 184

hours of activity data fiom 7 February to 10 March. Data were collected throughout the

study ma, fiom whichever telemetry station(s) had access to multiple radio-rnarked

Duniin at a given time.

The only other activity that caused the radio pulse rate to increase was downward

flight, so 1 devised a correction factor to separate flying h m foraging activity. The radio

puise rate doubled when Dunlin flew downward, but not when they flew horizontally or

upward. To determine how ofien elevated puise rates represented flying versus foraging

activity, and to determine the overall amount of time the radio-macked individuals spent

flying versus foraging in 1998, an observer noted whether or not each radio-tracked

individual was flying while collecting pulse rate data. This was possible because flocks

of Dunlin were easily observed in flight and the direction of the signal would move with

a fluck if the bird was flying. Elevated pulse rates atûibuted to flight were omitteâ h m

calculations of the amount of time individuals spent foraging in 1998, and the proportion

of elevated pulse rates atûibuted to flight was used to adjust the data collected in winter

and spnng 1995196. Flight could not be observed at night, but 1 could hear îhe radio

signal moving when birds were in flight, and these records of elevated puise rates were

deleted fiom the data sets used to calculate time spent foraging.

Feeding activity data were obtained fiom a total of 57 radio-marked Dunlin who

survived for a minimum of two weeks, three of which were deleted due to activity switch

maifunctions (invariant pulse rates even though the bird was dive and rnoving around).

Flying activity data were obtained fiom a total of 15 radio-marked Dunlin during 1998.

Mortality

When the activity switch indicated no changes in activity and the radio signal

indicated no movement over a number of tracking m s , efforts were made to retrieve the

radio guided by a receiver and hand-held H antenna. Bird remains found with radios

were confinned as cases of predation and the predator species identifieci where possible

(e.g., known roost sites of a particular predator). Radios that were retrieved h m tidal

rnudlfats, where remains would have washed away, and radios that could not be retneved,

but whose signais emanated fiom areas of known predator activity, were presumed to

have died. These radios may have fallen off the birds, however, no radios were found

without remains except for one in a tidal area. Six birds for which radio signais

disappeared f?om the area were also noted and considered as possible cases of mortality.

These birds were al1 located at least once and disappeared within the f%st week &er

banding, so radio failure was considered an unlikely explanation for the disappearance of

the signais, albeit not impossible.

Behavioural observation

The observer used binoculars and a timer to determine the amount of time Dunlin

flocks at Brunswick Point spent flying throughout the day and across tide stages in 1998.

Observations were made during 77 hous h m 9 February to 10 March 1998 fiom a

vehicle on the dyke. The observer recorded the amount of time that the bulk of the

Dunlin fiock(s) was(were) in flight. 1 added up the total number of hours in flight and

divided it by the number of hours of observation to obtain the mean percent of time in

flight, overall and by tide stage.

Scan and focal sarnpling were used to detemine how much time Dunlin spent on

each of the following activities: foraging, flying, preening, vigilance (high alertnesd neck

stretched), ruosting head up, rwsting head down, and agression. Samples were

collected fiom a vehicle on the dyke using a spotting scope and a micro-cassette tape

recorder. Al1 sarnpling took place during the day at Brunswick Point h m 7 February to

10 March 1998, in between the collection of telemetry activity samples. First, the

observer recorded the date, time, and macnhabitat (marine or terrestrial), and estimated

the size of the flock and its average density in bird lengths (the average # of bird lengths

between birds in the flock). For scan samples, the flock was then scanned and the

numbers of birds performing each of the behaviours described above were recorded. For

focal samples, mdomly selected individuals were observed for up to three minutes, and

their activities recorded in ceal time. Later, a timer was used to determine fiom the tapes

how long each bird was engaged in each behaviour.

SUMMARY STATISTICS AND ANALYSES

1 calculated the percentage of time each individual spent foraging by tide stage

(hi& and low), macro-habitat (marine and terrestrial), and time of day (day and night) in

the case of radio-marked individds. Al1 statisticai test results were considered to be

significant at P < 0.05, however, results with P-values between 0.05 and 0.1 are reporteci

as possibly significant biologically. Interaction tenns were considered to be statistically

significant at P c 0.10, since significance tests for interaction terms have lower power

124

than those for main effects (Littell et al. 1991). Results of 2 or 3-way ANOVAs reported

here were those of the reduced models in cases where the interaction term was not

significant, and 1 report Ieast-squares means taking the other factor(s) into account. Al1

percentage data that were tested siatistically were arcsine transformed and, shce sample

sizes varied among individuals, weighted by sample size (the number of activity data

points). Since some of the data were not tested statistically, I specifi the results of

statistical cornparisons (e.g., X was significantly larger than Y, as opposed to X was

larger than Y).

Randomization models for analyses using individual Dunlin

My samples of individual radio-marked Dunlin often did not meet the assumption

of normality, so 1 used randomization models to test the robustness of the results of al1

ANOVA and regression analyses described in this study. For example, to randomize the

results of a 2-way ANOVA testing for differences in the percent time spent foraging by

tirne olday and tide stage, the computer randornly shuffled the time of day and tide stage

designations of the birds in the data set and ran ANOVAs on 1OW samples generated in

this way. 1 then obtained a P-value by comparing the F-statistic obtained fiom the

ANOVA using the actual time of day and tide stage assignments to the distribution of F-

values obtained from the randomly generated samples.

Time spent foraging

By telemety

1 used two-way repeated measures ANOVA to detennine whether there were

differences in the percentage of t h e spent foraging by tide stage or sampling period.

Tide stage (F,,= 154.6, P < 0.001) and sampling priod (F,,, = 23.5, P < 0.001) were

boih signifiant factors, so they were included in subsequent tests.

I used three-way repeated measures ANOVAs to detemiine whether there were

differences in the percentage of time Dunlin spent foraging separately by t h e of day, by

macfo-habitat, or by sex (each considered wiîh tide stage aMi sampling period). Dunlin

codd be coniidendy aged ody during the winter 199996 sampling period, so 1 used a

three-way repeated measures ANOVA to determine whether there were differences in the

percentage of t h e spent foraging by age class (considered with tide stage and sex) within

that period. To examine the relative importance of day and night foraging to individuals

of different age and sex classes, 1 used four-way (sex and time of day with tide stage and

sampling period) and three-way (age and time of day with tide stage within the winter

1995196 sampling period) repeated measures ANOVAs.

To determine whether foraging time differed by season andor year, 1 broke the

data down into two separate data sets. 1 tested for a seasonal effect by comparing data

from winter 1995196 with data fiom spring 1996, and 1 tested for a year effect by

comparing data fiom spring 1996 with data fiom spring 1998 using separate 3-way

ANOVAs (considered with tide stage and sex). Within the spring 1996 sampling period,

Dunlin were captured at two separate sites (WI and BB), so 1 first had to determine

whether data fiom the two sites could be pooled. 1 used three-way repeated measures

ANOVA to test for differences in foraging activity between the sites (considered with

tide stage and sex). Percent time spent foraging did not differ between WI and BB (FI.,, =

0.3, P = 0.66), so data from these two sites were pooled to represent the spring 1996

sampling period. 1 dso used three-way repeated measures ANOVAs to examine the

relative importance of &y and night foraging to individuals in different seasons (season

and time of day with tide stage).

1 calculated the mean number of hours that Dunlin spent foraging each day, by

macro-habitat, time of day, and sarnpling period (winter 1995196, spring 1996, and spring

1998). 1 did this by fkst multiplying the average number of daytime hours (one half hour

before to one half hour after sunrise and sunset) and nighttime hours in each of the three

sampling periods by the mean within-bird percentages of time Dunlin were present in

each macro-habitat during day and night (Chapter 2). 1 then mdtiplied the result by the

mean within-bird percentages of t h e they spent foraging in each macro-habitat during

the day and at night. We did not obtain any telemetry data on the percentage of time

spent foraging by Dunlin using the terrestrial macro-habitat during the &y, so instead we

used the data obtained by observation (scan sampling). Since only one such sarnple was

obtained, no standard errors were calculated for this category. The number of day plus

night hours within each sampling period and macro-habitat summed to 24, and the

percentages of time Duniin were present in each of the two macm-habitats within each

sampling period and time of day sumrned to 100%.

By observation

The nurnbers of Dunlin in a flock that were foraging (scan samples) and the

amounts of tirne that individuals foraged (focal samples) were converied into percentages

and summarized (using means and standard errors) separately by tide stage for

comparison with the radio telemetry estimates of percent time spent foraging.

Time spent flying

By telemetry

1 calculated the percentage of time each individual spent flying in spring 1998 by

tide stage during the day, and weighted the percentages by the number of observations

obtained. 1 caiculated the mean number of hours that Dunlin spent flying each day by

multiplying the nurnber of daylight hours by the mean within-bird percentage of time

they spent flying. Finaily, 1 used a two-way repeated rneasures ANOVA to compare

percent time spent flying by tide stage and sex. Dunlin could not confidently be aged in

spring so no age class comparison was made.

By observation

1 calculated the total number of h o u the Dunlin flocks were in flight divided by

the total observation hours (77), overall and by tide stage, for comparison with the radio

telemetry estimates of percent time spent flying.

Time spent in other activities

The nurnbers of Dunlin in a flock performing each of the behaviours described

above (scan samples) and the amounts of tirne individual Dunlin spent performing each

activity (focal samples) were converted into percentages. These, dong with flock sizes

and densities were summarized (using means and standard errors) separately by tide stage

and macro-habitat.

Body size

Dunlin are sexually size-dimorphic, with females, on average, larger than males. I

therefoce examined whether the male and female Dunlin used in my sample differed

significantiy in body size. To obtain an overall measure of body size, 1 used principal

components analysis and data on culmen length, wing length, and weight (Rising and

Somers 1989). The first principal component eigenvectors had similar signs and

magnitudes for each measure (0.566.59), and the cumulative correlation matrix

eigenvalue was 0.68, so PCI was used as a measure of body size. 1 used two-way

ANOVA and Bonferroni adjusted multiple t-tests to compare PC 1 by sex and sarnpling

period. PCI for each individual was weighted by the telemetry activity sample size. 1

also used two-way ANOVA and Bonferroni adjusted multiple t-tests to separately

compare each of the three M y size variables (weight, culmen length, wing length) by

sex and sampling period. One Dunlin h m the spring 1998 sarnpling period was deleted

due to missing body size data.

Mortaliîy

1 calculated the percent of radio-marked Dunlin confirmed dead, presurned dead,

or whose signals disappeared (see mehods for definitions), separately by sex.

RESULTS

Time spent foraging

Radio-marked individual Dunlin spent on average at least 15.7 hour per 24-hour

day foraging (Table 5.1). They spent a significantly higher percentage of their t h e

foraging during low than high tides IF,$,= 154.6, P < 0.00 l), a difference of

approximately 30% (Table 5.2). The percentage of t h e that Dunlin spent foraging did

not dXer significantiy between day and night or between marine and terrestrial macro-

habitats (Tables 5.3 and 5.4). Since D d i n spent littie t h e in terrestrial habitat during

the &y (Chapter 2), the total number of hours that Dunlin spent foraging in the marine

habitat was greater than in the terrestrial habitat (Table 5.1).

Dunlin spent a significantly higher percentage of their time foraging in spring

than in winter (1995196 only) both during the &y and at night (Table 5.3 and 5.4).

However, there was a significant interaction between time of day and season (Table 5.9,

so Dunlin spent a relatively greater percentage of their t h e foraging at night in winter

than in spring (Table 5.4). The total number of hours that Dunlin spent foraging was also

greater at night during winter, and greater during the day in spring (Table 5.1). The

difference between seasons in the relative number of foraging hours by time of day was

partly because there were, on average, 9.5 hours of daylight during the winter period of

the study, and 12.5 hours of daylight during the spring period. Dunlin spent significantly

more of their time foraging during 1996 than during 1998 (spring only) (Tables 5.2 and

5.3).

There was no statistical difierence between adult and juvenile Dunlin in the

percentage of time spent foraging (Tables 5.3 and 5.6), but male Duniin spent

significantly more time foraging than female Dunlin (Tables 5.3 and 5.7). There were no

significant interactions between time of day and age class when considered with tide

stage, or between time of day and sex class when considered with tide stage and sampling

period (Table 5.5).

The estimate of the mean percentage of time that radio-marked individual Dunlin

spent foraging during high tide was within less thm one percentage point of the mean

obtained by scan sarnpling (Table 5.8). These estimates were approximately 10 percent

lower than the estimate obtained by focal sampling. During low tide, the mean for radio-

marked individuals was approximately 12 percentage points higher than the means

obtained by scan and focal sampling.

Time spent Wing

Radio-marked individual Duniin spent on average 24% of their rime flying during

the &y in spring 1998, which translates into an estimated 3.0 hours daily, (Table 5.9).

The estirnates of mean percent time during high tide that radio-marked individuals

spent flying was within less than one percentage point of the mean percent tirne that

Dunlin Bocks were observed to spend in flight (Table 5.10). Dwing low tide, the mean

for radio-marked individuals was four percentage points higher than the observed flock

mean.

Dunlin spent significantly more tirne in flight during high than low tides (F,,,,=

45.8, P < 0.001) (Table 5.10). The sexes did not differ significantly in the percentage of

time ihey spent in flight (FI,,, = 1.6, P = 0.27) (Table S. 10).

Three hours per day in flight is a minimum value, since 1 was unable to observe

flying at night. However, at night, Dunlin were more dispersed, they did not engage in

high tide flock-flying, and they made significantly fewer (P < 0.001) shortdistance lateral

movernents (0.1-1 .O km) (Chapter 2, Table 2.6) than they did during the day. There may

aiso be seasonai or yearly variation in time spent flying, but 1 collected comprehensive

flight data only during the spring of 1998.

Time spent in other activities

Unmarked Dunlin, as was uue of radio-marked Dunlin, spent most of their time

foraging during both high and low tides (Tables 5.1 1 and 5.12). They spent less than haIf

a percent of their time engaged in aggressive behaviour (Tables 5.1 f and 5.12).

lndividuals spent more time roosting head down and preening during high than low tide.

Only one flock of an estimated 300 birds was sampled in the terrestrial (field) macro-

habitat, and while there, most tirne was spent flying, followed by foraging, roosting with

their heads up, and vigilant (Table 5.12). No time was spent preening or k ing

aggressive, and less than half a percent of their time was spent roosting with their heads

dom. Flock density was lower during low than high tides.

Body sue

The fint principal component (PC 1) differed significantly by sex (F,,, = 47.3, P =

0.001) with average females k ing larger than average males (Table 5.13). Examining

the three body size variables separately (weight, culmen length, wing length), al1 were

larger for females than males (F,,4,> 10.6, P > 0.01). PCl did not differ significantly

among sampling periods (F2,4,= 4.5, P = 0.05), but the winter 1995196 and spring 1998

p u p s differed significantly by Bonferroni adjusted multiple t tests. Examining the three

body size variables separately, only weight differed significantly among sampling periods

(weight: F,,= 1 1.8, P < 0.001; culmen and wing lengths: F,,4, < 2.1, P > 0,24), because

of the significantly higher weight exhibited by Dunlin in the winter sampling period than

by Dunlin in either of the spring sampling pends (Bonferroni adjusted multiple t-tests).

Mortaliîy

Of the conf'umed mortalities, 3 were found amid seved other Dunlin carcasses at

Peregrine Falcon roost sites, 2 were found in areas frequently used by roosting falçons

and Short-eared ûwls (Asioj7ammeus), and one was found in a pile of owl pellets under a

Barn ûwl (Tyto a h ) rwst site. Of the presumed mortalities, two occumd out in

Brunswick Point marsh and could not be retneved due to rough terrain. Both of these

signais emanated from areas frequently used by roosting faicons and Short-eared Owls.

The third preswned mortality was a radio retrieved h m the mudfiat less than 50 m from

the dy ke near Mud Bay. The radio was found in an area subject to tidai action where

remains wodd likely have floateâ away.

There was no statistically significant difference between maies and females in the

numkr confirmed or preswned dead = 0.1, P = 0.72), or in the number that

disappeared fiom the study area (x', = 1 -9, P = 0.1 7) (Table 5.14).

DISCUSSION

individual Dunlin in the Fraser River Delta spent on average at least 15.7 hours

per 24-hour day foraging (dependhg on season), and another 3 hours per day flying

(Tables 5.1 and 5.9). This lefi at most 5.3 hours per day for other activities such as

roostiag, preening, vigilance, and aggression (Tables 5.1 1 and 5.12). Dunlin wintering in

Briîain, approximately 4 O latitude M e r north than those in the Fraser Delta, experience

the coldest winter weather of any Dunlin (Summers et al. 1998), but spent only 9.5 to I 1

131

hows per day foraging (Goss-Custard et al., unpubl. data in Townshend 198 1). The

difference in foraging t h e between Dunlin in Britain and in the Fraser Delta may be due

in part to differences in disturbance by predators, since the 3 hours Duniin in the Delta

spent flying each day was likely in response to predation risk (see below).

Time-activiîy budgets by time of day

The percentage of t h e that individual Dunlin in the Fraser Delta spent foraging

did not differ significantly by time of day (Tables 5.3 and 5.4). The total nurnber of

hours spent foraging exceeded the nurnber of available daylight hom, so it is unlikely

that Dunlin could have met their energetic requirements without foraging both dimally

and noctumally. However, the data do not indicate a clear choice for day or for night, so

neither the "choice for day" nor the "choice for night" hypotheses appear to explain the

foraging pattern of Durilin in the Delta.

Dunlin spent les than 79% of the low tide period foraging during winter

199511996 and spring 1998 (Table 5.2). They were not prevented fiom foraging by

inclement weather or high water levels during those times, so the "feed whenever

possible" hypothesis does not explain Dunlin foraging patterns in the Fraser Delta.

The fact that Dunlin spent a similar percentage of time feeding by day and by

night may provide support for the "feed day and night to avoid weight-related risk"

hypothesis. However, tfüs hypothesis is dificult to evaluate without knowing the relative

intake rates by day and night. In addition, Dunlin spent on average 3 hours flying during

the daytime, so the2 energetic expenditure may be greater during the day (even assuming

higher thermoregulatory costs at night). Finally, variation in M i n weights throughout

the 24-hour day may not be inconsistent with the "feed day and night to avoid weight-

related risk". If predation risk is higher during the day, for example, individuals may

chwse to maintain iower weights at that time.

The fact that Dunlin spent the same percentage of time feeding by day and by

night may a h appear to support the "indifference" hypothesis. However, this hypothesis

assumes that there is either no difference or a balance between day and night in the costs

and benefits of foraging, and such does not appear to be the case in the Fraser Delta (see

below).

Alternatively, one could simply hypothesize that individuals should forage dwing

the relative proportion of day and night hours that optimizes the state-dependant balance

between the risks of starvation and predation. For example, individuals wouid choose to

maxi& day foraging time and minimize night foraging time where 1) food intake rate

was higher during the day than at night and predation risk was equal or lower during the

&y than at night, or 2) predation risk was lower during the &y than at night and food

intake rate was equal or higher during the day than at night. Individuals may choose to

forage by both day and night where food intake rate and predation tisk are both higher

during the day or at night. For example, where both foud intake rate and predotion risk

are higher during the day than at night, individuals may choose to spend relatively more

time foraging during the day at times when they perceive themselves to be at high risk of

starvation, and relatively more time foraging at night at times when they perceive

themselves to be at low risk of stamation.

For Dunlin wintering in the Fraser River Delta, the evidence appears to indicate

that both food intake rate and predation risk may be higher during the day than at night.

ûunlin forage using both visual and tactile methods, and can switch between using a

more visual foraging mode during the &y to a more tactile mode at night (Evans 1987,

Mouritsen f 993 and 1994, Zwarts et al. 1990). Tactile foraging success should be

affectecl littk by light conditions, unless birds use sight to decide where to forage by

touch (Sitters 2000). Assuming qua1 prey density, foraging by exclusively tactile

mettiods could remit in lower intake rates for Dunlin, but the magnitude of this potential

difference has not yet ken determined. The difference may not be substantial,

particularly on moonlit nigbts, which is a preferred time for nocnumal foraging in

California Min @odd and Colwell 1998), or in the terrestrial habitat, where there can

be a considerable amount of man-made illumination at night. In a rare quantification of

relative day and night energy intake by visualty-foraging shorebis, both Grey Plovers

(Pluvialis squatamla) and Whimbrels (Numenius phaeopus) in South Afnca foraged

more slowly at night, but did not differ significantly in energy intake rates by tirne of &y

(Twpie and Hockey 1993). However, Dunlin can use visual cues to locate prey during

the day, thereby lowering their search time by up to 40% (Evans 1986), so food intake

rate may be higher during the day than at night.

Some invertebrate species have been found to increase their availability to

shorebirds at night by moving up the sediment column or becoming more active (Dugan

1 98 1, Townsend et al. 1984, Evans 1987, Robert and McNeil 1989), but 1 have no

convincing evidence that marine prey availability varies by time of day in the Fraser

Delta. My limited data show that large annelids may be more available (P = 0.05) but

crustaceans may be less available (P = 0.06) at night than they were during the day

(Chapter 2, Appendix, Table A.2.3).

Dunlin in the Fraser Delta are preyed upon by both diumal and nocnirnal raptor

species (this study), however, shorebirds were negligible contributors (less than one

percent) to the diet of the most cornmon nocturnai raptor species in the study area (Barn

Owl) (Dawe et ai. 1978, Campbell et al. 1987). Dunlin may aiso be taken by gulls and by

rats on occasion (D. Dekker and L. Evans-Ogden, pers. comrn.). Of the four confinned

kills of radio-marked Dunlin (radios found with remains) for which the predator species

could be identified, three were taken by diurnally active Peregrine Falcons and one was

taken by a noçtumally active Barn Owl. The two other confirmed kills were found in

areas where fdcons and Short-eared Owls (a diumaily and noctunially active owl

species) were hquentiy sighted. If we use the data on the four predators identified to

species as an index of the relative risks posed to Dunlin by diurnai and nocturnal

predators, in addition to the fmding that Dunlin are rarely taken by Barn Owls in the

study area (Dawe et al. 1978, Campbell et ai. 1987), then the risk of predation wouid

appear to be higher during the day than at night.

Time-activity budgets by macro-habitat

Duniin in the Fraser River Delta may be using a cost-benefit analysis of the trade-

offs between acquiring food and avoiding predation to choose not only which time of &y

to forage, but also in which of the two macro-habitats to forage. The percentage of tirne

Dunlin spent foraging did not diier significantly between macro-habitats, however, uiis

does not mean that the two rnacro-habitats were equally important to D d i . The

terrestrial habitat was rarely used during daylight (Chapter 2). On average, Dunlin spent

between up to 14.0 hours per 24-hour day foraging in the marine habitat (winter and

spring, respectively), and up to 4.0 hours foraging in the terrestriai habitat (winter and

spring, respectively). In addition, there was considerable individual variation in the use

of the terrestrial habitat at night, with some Dunlin aiways located and some never

located in the terrestrial zone (Chapter 2). Some, but not dl, Dunlin in the Fraser Delta

may therefore not have k e n able to meet their energetic requirements without the

availability of the terrestrial macro-habitat as an alternative foraging site.

The risk of predation may be higher in the terrestrial habitat because Dunlin are

closer to raptor roosting and hiding places from which surprise attacks can be launched,

and they must contend with greater density and diversity of predator species (Whitfield

1985, Butler 1999, Caldow et al. 1999). The marine macro-habitat is flac open, and two-

dimensional, allowing for early detection of attackers, whereas terrestrial macro-habitat is

three-dimensional, 6 t h hedgerows and trees distributed throughout. Furthet evidence to

support the hypothesis that predation risk is higher in the terrestrial macro-habitat can be

found in the time-activity budgets obtained by scan sampling (Table 5.12).

Unfortunately, the terrestrial sample size was just a single flock, since the birds rarely

used his macro-habitat during the day, so these data must be viewed with caution.

Dunlin in the terrestrial macro-habitat did not preen or exhibit aggressive behaviour, and

spent less than half a percentage of their time roosting with their heads down, al1 postures

that might distract their attention fiom oncoming predators (Redpath 1988). These same

behaviours together consumed 29.6% of the time for Dunlin in the marine habitat.

W i n in the terrestrial habitat also spent a relatively higher percentage of their tirne

vigilant and roosting with theù heads up than birds in marine habitat, both postures in

which bu& can keep an eye out for attacking predators. Turnstones (Arenario interpres)

and Purple Sandpipers (Calidris maritirna) increased their levels of vigilance with

decreasing visibility due to obstructions on the landscape (such as boulders, wrack banks)

(Metcalfe 1984).

Predation risk would appear to be highest in the terrestriai habitat during the day,

and lowest in the marine habitat at ni@. In comparing the terrestrial habitat at night to

the marine habitat during the day, relative risk would depend on the relative importance

of time of day and macro-habitat in deterrnining the level of risk. 1 do not have any data

on the relative food intake by macro-habitat or by time of day, so 1 cannot directly

determine the relative benefits of foraging in each time/place. Use of the terrestriai

habitat may ailow for continued or increased food intake during high tides when marine

habitat is less available or altogether submerged, or during periods of rainfall, when

fieshwater input may decrease the availability of marine invertebrates, and increase the

availability of terrestriai invertebrates (Goss-Custard 1970b, Gerstenberg 1979, Goss-

Custard 1984, Pienkowski 198 1, Heitmeyer et al. 1989). Ongoing research estimates the

contribution of the terrestrial macro-habitat to average approximately 30% of the diet,

with a range of 1-92% (L. Evans-Ogden, pers comm.). This extensive range coincides

with the telemetry data showing that the relative use of the terrestriai habitat varies

among individuals. To some Dunlin it may be crucial, while to others, it may not be

required. Among groups of Duniin in the Fraser Delta, the relative use made of the

terrestrial habitat did not differ significantly between age classes (FI,,, = 0.5, P = 0.51,

Chapter 2), but there was some indication that it may have ken used more by Boundary

Bay birds than by birds fiom the other two sites (F,>, = 4.4, P = 0.06, Chapter 2, Table

2.4). There was also some indication that there may have been a negative relationship

between culmen length, a continuous measure of sex ratio, and the percent of locations in

terrestrial habitat (F,,, = 3.7, ? = 0.10, P = 0.06, Chapter 3).

Dunlin were rarely located in terrestrial habitat during the day, the potentially

riskiest time/place combination, and use of the terrestrial habitat at night was significantly

greater during high tide. However, some Dunlin did use the terrestriai habitat during low

tide at night. Assuming that for a given individual, intake rate would be higher in the

terrestriai than in the marine habitat at night, it might choose to take on the greater risk

associated with foraging in terrestrial habitat if its interna1 state (fat stores) indicated it

was at greater risk of starvation.

High tide flock-t'tying behaviour

Dunlin in the Fraser Delta spent on average 3 hours flying during daytime each

day, sometimes while mudflat foraging habitat was available. This flying behaviour is

consistent with the hypothesis that during times of elevated risk, such as daytime high

tide, when Dunlin are pushed closer to the terrestrial habitat, they benefit from flying

instead of foraging. Dunlin may choose to forego foraging during daytirne high tides and

instead spend more time foraging at night. Flock-flying behaviour, although

energetically expensive (Castro and Myers 1989, Hotker 2000), can decrease the risk of

predation by avian predators (Brennan et al. 1985, Whitfield 1985, Cresswell 1994,

Dekker 1998,1999, Hotker 200). The risk of predation is thought to be higher closer to

the terrestrial habitat because Dunlin are closer to raptor roosting and hiding places, and

may be more easily surpriseci due to restricted visuai horizons relative to completely open

mudflats (Whitfield 1985). Peregrines hunting Dunlin in the Fraser Delta were successful

on 33% of surprise attacks and only 8% of aerial chases (Dekker 1998 and 1999), and

other studies have shown hat fdcons hunting Dunlin had the greatest success when they

attacked by surprise (Page and Whiteacre 1975, Whitfield 1985, Cresswell 1996). Hotker

(2000) found that flock-flying by Dunlins in the Wadden Sea occurred at only one of 10

roost sites. The site in question was the only one lacking an area of sparse or low marsh

vegetation to allow for early predator detection, and there were taIl trees suitable for

concealing predators in closer proximity io the roost than at any other site. Moreover,

flock-flying at this site was observed on only 5 of 35 occasions, al1 of which occurred

afier the anival of migrant Peregrine Falcons and Sparrowhawks (Accipiter nisus) in rnid-

September.

Time-activity budgets by season

Dunlin spent more of theù iime foraging in spring than in winter (Tables 5.1 and

5.2). An increase in foraging îime c m arise kom changes in either energetic

requirements or intake rates (Zwarts et al. 1990). In spring, the accumulation of fat for

migration to the breeding grounds and the partial pre-alternate molt require additional

energy (Myers and McCaffrey (1984), Zwarts et al. 1990, Morrier and McNeil 199 1). At

the same time, Dunlin are subject to higher temperatures, reducing the energetic costs of

thermoregulation, although Dunlin living under thennoneutral conditions in Mauritania

also spent up to 30% more time feeding in March and Apd than in February (Zwarts et

al. 1990). The spting decrease in invertebrates (McEwan and Fan 1986) could have

contributed to the increase in time spent foraging. However, Dunlin weights are

increasing at that time (Kaiser and Gillingham 198 1, McEwan and Whitehead 1984), and

there is still suficient invertebrate biomass to allow hundreds of thousands of migrant

Western Sandpipers (Calidris mauri), which feed on many of the same invertebrate prey

items (B. Elner, pers. comm.) to accumulate the fat necessary for the next leg of their

northward migration.

Dunlin in the Fraser Delta lose weight (approximately 4 g) in late winter (January-

February), and then accumulate that much and more in preparation for spring migration

(February-April) (Kaiser and Gillingham 198 1, McEwan and Whitehead 1984). It has

been hypothesized that the winter weight loss may have been due to a decrease in

available invertebrates (Pienkowski et al. 1979, Kaiser and Gillingham 198 1, McEwan

and Whitehead 1984). Although the invertebrates may have been less active at that time

due to Iowa temperatures (thereby providing fewer visual cues to their presence), they

occurred in the top 5 cm (within the Dunlin bill range) of the sediments in higher

densities in winter than in spring (McEwan and Fan 1986). If the birds were losing

weight because their intake rates were declining, one solution would have been to

increase their time spent foraging. However, Dunlin spent less time foraging in winter

than in spring. An alternative hypothesis to explain the variation in weights through the

non-breeding season is that the changes might be due to changes in the relative risks of

starvation and predation (Lima 1986). The likelihood of a bad weather event decreases as

winter progresses, so Dunlin could be choosing to forage less and to use up some of their

fat reserves, thereby making themselves more aerodynarnic and improving theu ability to

evade predation (Metcalfe and Ure 1995). The mean weight of Great Tits (Parus major)

in England varied inversely with the size of their avian predator population, and 90% of

the annual change in mean weight was attributed to adjustments made by individuals

(Gosler et al. 1995).

In winter, Dunlin spent less time foraging dwing daj-time tow tides (71% of their

tirne) than they did in spring (97% of their time), There was also a significant interaction

between time of day and season in the percentage of time individual Dunlin spent

foraging (Table 5.5). Dunlin spent a relatively greater percentage of their time foraging

at night in winter than in spring (Table 5.4). These findings are consistent with the

hypothesis that during times of elevated risk, such as when they are carrying more fat and

may be l e s maneuverable, Dunlin may choose to spend more time foraging at night,

when predation risk is lower (see above). Dunlin carry more weight during the period

designated as 'winter' in this study than they do dwing the period designated as 'spring'

(Table 5.13) (Kaiser and Gillingham 198 1, McEwan and Whitehead 1984). Dunlin begin

to accumulate fat for migration in 'spring' (Febniary and March), but they do not reach

the level of their 'winter' weight until April (Kaiser and Gillingham 1981, McEwan and

Whitehead 1984).

Time-activity budgets by year

Dunlin spent a higher percentage of their time foraging in 1996 than 1998 (spring

only), a difference of 1.5 hours per day on avetage (Tables 5.1 and 5.2). Nineteen ninety-

eight was an el nia0 year, and in California, the resulting increase in tide levels caused a

considerable decrease in the avaiiability of intertidal habitats. In the Fraser Delta, the

actual mean tide level was only 0.2 m higher in 1998 than it was in 1996 (Institute for

Ocean Sciences, pers. comrn.), so the el nifio event did not cause a large decrease in

available intertidal foraging habitat, and consequently foraging tirne, at tfijs site. Mean

body weight and structural size (culmen and wing) did not differ between the two spring

sampling periods (Table 5.13), so the lower foraging time of Dunlin in 1998 did not

appear to result in a decline in boày condition. The average spring temperature was

slightly higher in 1998, so it is possible that thermoregulation costs were lower that year,

allowing for a decrease in the t h e spent acquiring energy.

Time-activity budgets by age and sex class

Adult and juvenile D d i n did not differ in the percentage of time they spent

foraging, nor did juveniles spend relatively more or less time foraging at night than adults

(Tables 5.3 and 5.5). It appears, therefore, that the juveniles either foraged as efficiently

as adults, or expended less eaergy. Female Dunlin spent less time foraging than males,

even though they were larger (Table S. 13). Assuming that both sexes were either meeting

their energy requirements, or losing and gaining weight at the same rate (Kaiser and

Gillingham 1985), females must either have been consuming more, larger, or better

quality food per unit time. Females did not spend less time in flight than males (Table

5.12), and the= was some indication that they may have been less likely to be located in

the terrestrial macro-habitat (Chapter 2) where winds, and therefore thermoregulation

costs, were probably lower (due to landscape features). Therefore females do not appear

to have used behaviowal mechanisrns that lowered their ovedl energetic requirements

compared to males, nor is there any evidence that energy assimilation eficiency is any

better in females (C. Guglielmo, pers. comm.). it is possible that femaie Dunlin might

have higher prey capture success rates than males because they have longer culmens and

may therefore be able to reach intertidal prey items that males c m o t access. Female

Curlew Sandpipers (Calidris ferrugineu) and Bar-tailed Godwits (Limosa lapponica) had

greater foraging success and foraged faster (higher probe rate) than males (Smith and

Evans 1973, Puttick 198 1). The average bill length of female Curlew Sandpipers was 4

mm longer than that of males, approximately equal to the size difference in the mdio-

marked Duniin in this study.

Advantages of collecting timeaetivity data using radio telemetry

The differences between the mean percent time that radio-marked and unmarked

Dunlin spent foraging provide examples of how the limitations of observational studies

can bias results. Whittingham (1996) found that radio activity switches were more

accurate than focal sampling at estirnating the peck-rate of foraging European Golden-

Plovers (Pluviulis apricaria), and hypothesized that this was because of an increased

probability of observers failing to record pecks when peck rate was high. In this study,

the difference in foraging t h e between radiemarked individuals and unmarked

individuals used for focal sampling (Tables 5.8 and 5.9) may have been due in part to an

inability to follow unmarked individuals that made more than short flights. The

percentages of time spent in each behaviour af5ected the values for the other behaviours,

so my inability to follow unmarked individuals that moved could bias my time-activity

samples by decreasing the time individuals were observed to be flying and thereby

increasing the apparent percentage of time spent foraging and in other activities. In

comparing the radio-marked birds to flocks sampled by scan sampling, there was no

difference in time spent foraging at high tide (Table 5.8). During low tide, the differences

between radio-marked birds and unmarked birds sampled by both focal and scan

sampling techniques may have ken due in part to the fact that the visual observations

were limited to birds occurring less than 500 m of the dyke, whereas the radio-marked

individuals could be monitored as they rnoved downshore with the receding tide. There

was no difference in the mean percent time that radio-marked individuals and Dunlin

flocks spent flying during high tide (Table 5.9). During low tide, the value obtained by

observation was four percent lower than that obtained by telemetry, perhaps in part

because the birds were further h m the dyke at that tirne and harder to keep track of by

observation iilone.

Conclusions

Dunlin in the Fraser River Delta, at the northem end of their non-breeding

distribution, are subject to high thermoregulatory costs. In addition, they spent on

average 3 hours per day in flight, as a tactic to decrease the risk of predation by faicons.

To meet theu energetic costs, Dunlin spent more than 60% of their time foraging, both

day and night, and in both marine and tenestria1 habitats (Table 5.4). Aside fiom

foraging and flying, Dunlin had a maximum of 5.3 hom pet 24-hour day for ail other

activities such as roosting, preenhg, vigilance, and aggression (Tables 5.1 1 and 5.12).

Thermoregulatory costs, predation risk, and a lack of suitable (agriculturai) terrestrial

habitat adjacent to the marine habitat may together explain the absence of any large

numbers of Dunlin north of the Fraser Delta in winter.

Table 5.2. Least squares mean percent time Dunlin spent foraging (+ SE) by tide stage within each of the three sampling periods. N = number of individuals.

Winter 1995196 N Spring 1996 N Spring 1998 N

Tide Stage High: 48.6 + 2.8 2 1 66.6 2 3.4 16 46.8 2 2.9 16 LOW: 77.1 + 2.9 20 94.1 1: 3.7 16 78.1 + 2.3 16

Table 5.3 stage and sampling

. The effects on the percentage of time Dunlin spent foraging of time of day and macro-habitat (3-way ANOVAs with tide sainpling period), season and year (3-way ANOVAs with tide stage and sex), sex (3-way ANOVAs with tide stage and period), and age (3-way ANOVA with tide stage and sex within the winter 1995/96 sampling period).

Factors F d f P Data subsets used in analyses

Time of day O. 1 1,43 0.84

Macro-habitat 0.2 1,26 0.72

Season 48.4 1,33 < 0,001 Winter 1995/96 and spring 1996 only

Year 47.8 1,29 < 0.001 Spring 1996 and spring 1998 only

Sex 8.4 1,48 0.009

AlF 0.7 1,17 0,45

Table 5.4. Least squares rnean percent time Dunlin spent foraging (+ SE) by titne of day and by macro-habitat within each of the three sampting periods, taking tide stage into account. N = nurnber of individuais.

Winter 1 W5/96 N S~rina 1996 N S~r ing 1998 N

Time of Day Day : 59.6 2 2.7 22 82.7 2 2.9 16 63.7 5 2.1 16 Nighr : 65.6 + 2.9 2 1 74.8 + 5.1 1 1 59.2 + 3.6 12

Macro-habitat Marine: 62.3 + 2.4 2 2 83.2 + 2.7 16 63.1 + 2.0 16 Terrestrial: 63.5 2 4.0 15 70.5 + 6.6 5 60.6 & 6.1 7

Table 5.5. Results of four- and three-way repeated mesures ANOVAs examining interactions between tirne oïday and sex class, tirne of day and age class, and time of day and season in the percentage of tirne Dunlin spent foraging.

Numerator Denominaior Test Factor d f d f F-value P

3-way ANOVA Time of day *Season (time of day and season, Time of day*Tide stage*Season with tide stage)

3-way ANOVA Tirne of day*Age (tirne of day and age class, Time of day*Tide stage*Age with tide stage)

4-way Time of day*Sex 1 3 6 0.5 0.55 ANOVA (tirne of day Tirne of day*Tide stage*Sex 1 19 2.2 O. 17 and sex class, with tide stage, Time of day*Sampling period*Sex - 3 3 6 O. 1 0.9 1 and sarnpling period) Tinie of day*Tide stage* Sarnpling period*Sex 2 19 1.2 0.42

Table 5.6. Least squares mean percent time adult and juvenile Dunlin spent foraging (+ SE) in winter 1995/96, by tide stage and taking sex into account. N = number o f individuals. -

% Time Feeding N

Table 5.7. Least squares mean percent time male and female Dunlin spent foraging (2 SE) by tide stage (H = high tide and L = low tide) within each of the three sarnpling periods. N = nurnber of individuals.

Winter 1995/96 N Spring 1996 N Sprinn 1998 N

Males H: 51.253.7 10 H: 70.3 3- 4.6 7 H: 49.9 2 3.6 9 L: 77.9+ 3.6 10 L: 95.7 +- 5.2 7 L: 82.1 2 2.8 9

Females H: 44.2 2 4.3 9 H: 62.7 2 4.8 9 H: 41.8 + 4.5 7 L: 77.6 + 5.0 8 L: 92.7+ 5.0 9 L: 7 1 . 7 2 3 3 7

Table 5.8. Mean percent lime that radio-marked individual Dunlin (by telemetry) and unmarked individual Dunlin (by focal sampling) spent foraging (2 SE), and mean percent of individuals in Dunlin flocks (by scan sampling) cngaged in foraging (2 SE), overall and by tide stage in 1998.

Radio-marked N Unmarked N N individuals (# individuals) individuals (# individuals) Flocks (# flocks)

High tide 46.8 + 2.9 16 56.02 5.8 4 1 46.7 5 6.4 34

Low tide 78.1 2 3.6 16 66.6 2 4.2 69 64.7 + 4.8 39

Table 5.10. Least squares mean percent time radio-marked male and female Dunlin spent flying SE) by tide stage in 1998. N = number of individuals.

% Time Flying N

Males

Females

Table 5.11. Activity budgets o f unmarked individual Dunlin showing the mean percent time ( i SE) spent in each activity (focal sampling). N = number o f individuals.

High Tide Low Tide Mudflat Mudflat (N = 41) (N = 69)

Feeding 56.0 + 5.8 66.6 + 4.2

Flying 13.8 + 3.8 17.1 + 3.2 Roosting head down 15.4 + 5.2 2.6 4 1.8

Roosting head up 6.3 _+ 1.5 9.5 + 2.2

Vigilant 2.6 2 0.9 1.2 0.4

Preening 5.9 4 2.1 2.8 5 1.4

Aggression 0.2 + 0.1 0.3 - 0.2

Flock density nieasure 1.6+0.1 4.7 + 2.4 (inter-bird lengths)

*Flock densiiy nieasure increases witli decreasing density

Table 5.13. Least squares mean body size variables, PCl s, and overall % tirne foraging (+ SE) for male and female Dunlin used in the foraging time analyses, by sarnpling period and sex. N = nuniber of individuals.

Weight Culmen Wing PC 1 Overall (g) (mm) (mm) (68% of var.) % time foraging N

Winter 1995/96 Males 52.8 + I .O 36.5 + 0.4 ** 120.6 + 1 .O ** -0.5 + 0.3 ** 64.7 + 2.9 11 Females 55.0+ 1.3A 40.5 i 0.5 A 124.92 1.2 A 1.5 + 0.4 A 58.3 + 3.5 9

Spring 1996 Males 47.02 1.4 ** 36.3 + 0.5 ** 120.22 1.3 ** -1.520.4 ** 81.5 + 3.8 7 Females 51.5 + 1.4" 41.6 + 0.5 A 125.4+ 1.3 A 1.3 k0 .4 AH 77.1 L3.8 9

Spring 1998 Males 47.7 t: 0.9 36.6 + 0.3 ** 120.2 f 0.8 -1.4 + 0.3 ** 71.1 ~ 2 . 5 8 Females 50.7 + 1.1 " 41.2 + 0.4 A 121.1 + l . O A 0.3 + 0.3 " 60.4 + 3.1 7

** sexes were significantly different for that variable (P c 0.01). sarnpling periods with shared letters were not significantly different from each other for that variable (Bonferroni adjusted

multiple t tests, P < 0.05).

CHGPTER SIX

SEX RATIOS OF DUNLIN WINTERING AT TWO LATITUDES ON THE

PACIFIC COAST

Shepherd, P. C. F., D. B. Lank, B. D. Smith, N. Wamock, G. W. Kaiser, and T. D.

Williams. 2001. Sex ratios of Dunlin wintering at two latitudes on the Pacific Coast.

Condor 1 O3:352-360.

Reprinted with permission (BCwper ûmithological Society).

ABSTRACT

Latitudii clines in sex ratio during the nonbreeding season occur in some

shorebirds of the Scolopacidae. We compared populations of nonbreeding D d i n

(Calidris alpinapaciJca) h m two latitudes dong the Pacific flyway: the Fraser River

Delta, British Columbia and Bolinas Lagoon, California, to detemine whether, and to

what degree, they exhibited sex ratios consistent with a latitudinal cline. Dunlin are

plumage monomorphic, so we used a maximum likelihood mode1 to estimate overall and

monthly sex ratios for each population based on culmen length distributions. Sex ratios

in the Fraser River Delta were corrected for sex differences in habitat use. Monthly sex

ratios were similar at the two sites but varied throughout the winter, likely reflecting

differences in seasonal movement patterns between the sexes. Both populations showed

an overall bias toward males (Fraser = 61% males, Bolinas = 65% males). Since there is

no evidence to support the possibility of a skew toward males in C. a. pacrjica as a

whole, our data are consistent with some form of latitudinal cline in the sex ratio of C. a.

puc$ca. However, additional data fiom the Oregon coast, southem California, and

Mexico are required to resolve this question. We also tested the hypothesis that mean

body size within each sex is larger at the higher-latitude site (Fraser River Delta), but this

did not appear to be the case.

INTRODUCTION

Latitudinal clines in sex ratio during the nonbreeding season are common in birds

(Ketterson and Nolan 1983, Belthoff and Gauthreaux 199 1). In species known to exhibit

the phenornenon, females, which are usually the smaller sex, typically overwinter farther

fiom the breeding grounds than males (Ketterson and Nolan 1976, 1979, Belthoff and

Gauthreaux 1991). Three principal hypotheses have been put forward to explain this

difference in winter range (reviewed by Ketterson and Nolan 1983). First, the "body-

size" hypothesis proposes that the larger sex could winter at higher latitudes due to

superior fasting endurance and thermal efficiency. Second, the "dominance" hypothesis

proposes that the dominant sex will migrate to "optimal" wintering habitats, and the

suMominant sex will be forced to either undergo longer and more costly migrations or

occupy less suitable habitats. Third, the "arrival-time" hypothesis proposes that the sex

that benefits most by aniving on the breeding grounds earlier will winter closer to the

breeding grounds.

Myen (1981) tested these hypotheses using data on shorebird species with

different patterns of winter distribution of the sexes, with both male and female-biased

sexual size dimorphism, and with different mating systems and patterns of amival on the

breeding grounds. Conclusions about the relative importance of each hypothesis can be

drawn if they lead to testable predictions, such as in those species in which males are

smaller than females, but precede them to the breeding grounds. Myers found that males

wintered north of females in al1 species for which males amived first on the breeding

grounds, regardless of which sex was larger. Latitudinal clines in sex ratio were not

observed in species for whom sexes anived on the breeding grounds together. Myers

concluded that information on breeding ground amival schedules was "both necessary

and sufficient" to predict whether and how sexes would segregate latitudinally.

The North Amencan Pacific Coast subspecies of Dunlin (Calidris alpina pacifica)

breeds in Alaska and winters in significant numbers fiom the Fraser River Delta in

southwestern British Columbia to southeni Baja California (Warnock and Gill 1996).

Estuaries in that range support tens of thousands of wintering and migrating Dunlin each

year (Paulson 1993, Page et ai. 1999). During the nonbreeding season, most Dunlin are

found in coastal and adjacent agricultural habitats, although some spend part or al1 of the

season inland in freshwater wetlands and agciculhual habitats in the Willamette Valley,

Oregon, and the Central Valley, Caiifomia (Strawh 1967, Shuford et al. 1998). Dunlin

males arrive first on the breeding grounds, defend breeding territories, and are srnaller

than females (Holmes 197 1, Wamoc k and Gill 1996, R. E. Gill, unpubl. data).

The arrival-tirne hypothesis predicts that male Dunlin will winter farther north

than females, whereas the body-size and dominance hypotheses make the opposite

prediction. If the pattern observed in other species by Myen (198 1) is general, we expect

that the Dunlin population wintering in the Fraser Delta will be male-biased relative to

more southerly sites. However, following the logic of the body-size and dominance

hypotheses, we might also expect to find within-sex differences in body size, with larger

birds wintering in the Fraser Delta. To our knowledge, the latter possibility has not

previously k e n examined in this context for shorebirds.

Information on population sex ratios at different wintering latitudes is needed to

test these predictions. Ideally, sex would be determined unequivocally using a molecular

sexing technique (Baker et ai. 1999), but the costs for large data sets are not trivial, and

appropriate samples are not available fiom long-term data sets originating before the

techniques were available. Sex of Dunlin, which are plumage monomorphic, is ofien

assigned based on culmen length (MacLean and Holmes 1971, Pienkowski and Dick

1975, Ptater et al. 1977). However, geographically distinct groups of Dunlin also Vary in

culrnen length (Wamock and Gill 1996, Engelmoer and Roselaar 1998). Differences in

size and, in particular, culmen size, have been used to identifi the breeding origin,

wintering sites, and migration routes of different groups and subspecies of Dunlin and

other shorebirds (Pienkowski and Dick 1975, Browning 1977, Engelmoer and Roselaar

1998). Therefore, the use of culmen lengîh criteria fiom one population to determine the

sex of individuals in another population could lead to errors in the assessrnent of sex

ratio. We studied two nonbreeding populations of C. a. pacifica: one near 49"N latitude

on the Fraser River Delta, British Columbia, and one near 38% latitude on Bolinas

Lagoon, California. We estimated the sex ratios of these two populations using a

maximum likelihood mixture anaiysis of culmen length distributions derived from locai

populations. We used the ratios estimated to determine whether, and to what degree, the

pattern was consistent with a latitudinal cline, and we investigated within-sex differences

in body size.

METHODS

Reference sample population

A reference sample of culrnen lengths for Dunlin of known sex was created fiom

M i n collected on the Fraser Delta mudfiats from November through March 1992-1 995

(n = 182). We measured culmen length ( h m the tip to the margin between mandible

and feathers at the center of the upper mandible) to the nearest 0.1 mm. Sex was

detemilneci by dissection. Page's (1974) data collected fiom California museum

specimens of Dunlin were used as the reference data for Bolinas Lagoon.

Mist-netted sample popuktions

Dunlin were captured at night in mist nets, banded, rneasured, and released in the

Fraser Delta fiom October through May 1978-1979 (n = 13 16), and in Bolinas fiom

October through April1979-1992 (n = 1 179), as part of two separate studies. Dunlin in

the Fraser Delta were captured on mudflats, while Dunlin in Bolinas were caught in tidal

marshes. We rneasured culmen length to the nearest 0.1 mm and inferred each bird's age

based on its plumage (Page 1974, Prater et al. 1977).

Sex ratio estimation

We usai a maximum likelihood mixture model (Schnute and Fournier 1980,

Smith and Jamieson 1989) to estimate the population sex ratio (proportion of males) in

the Fraser Delta and Bolinas during the entire nonbreeding season (October through

April), during winter (December through Febniary), and by month. We used culmen

length as the only discriminating criterion. Although wing length and weight have been

used to discriminate between sexes in C. a. pacifca (Brennan et ai. 1984), we chose not

to include these variables. Both weight and wing length are known to Vary seasonaily,

and wing length exhibits more variation than culmen length between live and museum

specimens (Pienkowski and Dick 1975, Prater et al. 1977, Atkinson et al. 1981, Freed et

ai. 1996). In tact, Greenwood (1979), examining pst-mortem shrinkage of Dunlin skins,

detected significant shrinkage in wing length and no change in culmen length.

Mixture analysis is routinely used in fisheries investigations to separate sex and

age classes (Schnute and Fournier 1980, Smith and Jamieson 1989, Smith and Botsford

1998). For our application, the distribution of culmen lengths in the population was

modeled as an overlapping mixture of normally distributed male and femaie culmen

lengths. Five parameters are required to describe the distribution: mean and SD of

culmen lengths for each sex, and the proportion of one sex in the distribution of culmen

lengths. Given a reference sample of individuals of known sex (e.g., by dissection), the

model estirnates the mean size and SD of males and femaies in the reference and

'ûnkn0wn7' own"p1es together, as well as the proportion of one sex.

We used the mixture model in two ways. For the Fraser Delta, al1 five parameters

were estirnated simdtaneously, using culmen length data h m both the reference sample

and the unknown sample. For Bolinas, we had available only the means, SDs and sample

sizes of a reference sample h m Caliûoniia museum specirnens (Page 1974). We

accepteà these as a reference since Greenwood (1 979) detecteà no pst-rnortem

shrinkage in the culmen lengths of Dunlin. Thus, we used a Bayesian approach by

formally including the uncertainty (SE) in the point estimates of the means and SDs for

the Bolinas population (Walpole et al. 1998).

We calculated asymptotic standard mors for dl estimates and report sex ratios as the

percent of males f SE.

We tested the goodness-of-fit of our predictions to sampled data using the chi-

square diagnostic and judged that five outliers be removed. We compared the estimated

sex ratios of the F m r Delta and Bolinas samples using z-tests, and we used chi-square

goodness-of-fit tests to determine whether the sex ratios were significantly different frorn

1:l.

Measurement error

Culmen length measurements Vary both within and among observers (Barrett et

al. 1989, Lougheed et al. 1991). Dunlin used in our study were measured by several

different observers, so to determine whether there might be biases among sarnples, the

culmen lengths of eight Dunlin were measured three times each by four of the observers

involved in collecting the data. We tested for measurement variation among samples

using nested ANOVA, and calculated the overall percent measwement error (% ME) 2 using Bailey and Binid (1 WO) fomiula (Y. ME = 100% [ s ~ ~ ~ ~ ~ R / ( s ~ ~ ~ ~ ~ ~ + s

There was significant variation in culmen length rneasurements between the

observers that measured the Fraser Delta refemnce and mist-netted samples (0.3 mm,

F832 = 6.5, P < 0.001), with one always recording smaller measurements than the other.

Therefore we applied a correction factor to the Fraser Delta data. Culmen length

measurements were not significantly different between the Fraser Delta and Bolinas

observers (F8,8~ = 1.5, P = 0.2), and the overalI measurement error was 0.7%.

Capture bias

Sex differences in habitat use have been reported for severai species of shorebirds

(Townstiend 198 1, Puttick 1984, Warnock 1994, McCloskey and Thompson 2000). Sex-

specific habitat-use data were collected in the Fraser Delta using radio-telemetry during

the 1995-1996 n u n b d i n g season (P. Shepherd, unpubl. data). Both male and female

Dunlin use agricultural habitats adjacent to the mudfiats in the Fraser Delta,

predominanîiy at night when predation risk likely decreases; however, mdes appear to

use agricultural habitats more oAen than females (P.Shepherd, unpubl. data). There was

a marginally significant correlation between culmen length and the percent of time

Duniin were absent h m the mudflats at night (r = -0.3 1, n = 37, P = 0.06), with male

(shorter-billed) Dunlin absent 15.1 f 7.5% more often than femaie (longer-billed) Dunlin

(taking tide stage into account). Since the mist-netting for our Fraser Delta sample took

place at night on the mudflats, we corrected the estimated sex ratio. We did this by

multiplying the percentage of males by Il 5.1% and adding the variance in this correction

factor (7.5%) to the total variance of the percentage of males. Warnock et ai. (1995) also

found some habitat segregation between the sexes in Bolinas, but it was on a seasonal

rather than a daily scale and is accounted for in the month-to-month sex ratio data

reported below. Therefore no correction was made to the Bolinas data.

M y size

Culmen length is highly correlated with ovedl body size in Dunlin (P < 0.001),

so it was used as an index of body size (Engelmm and Roselaar 1998, P. Shepherd,

unpubl. data). ANOVA was used to compare the culmen length of the Fraser Delta

refetence sarnple for each sex by season: fail (October and November), winter (December

through February), and spring (March and April). No seasonal differences were found

for either males (F2,6Z = 0.1, P = 0.9) or fernales (F2, 14 =1.4, P = 0.3), so the sampies

were pooled for Mer analyses. Culmen lengths of adult and juvenile Dunlin were

compared, separately by sex, using t-tests. No age differences were found for either

males (f16 = -0.5, P = 0.7) or femaies (rI9 = -0.1, P = 0.9), so age classes were pooled.

We used one-tailed z-tests to determine whether the estimated mean culmen lengths of

M i n in the Fraser Delta were larger than those in Bolhas, separately by sex. We

report mean culmen lengths f SE below.

RESULTS

Sex ratios

The mixture model estimated the overall (October through Apcil) nonbreeding

percentage of males in the Fraser River Delta to be 53 t 3% (Fig. 6.1). When corrected

for sex differences in habitat use, the percentage of males in the Fraser Delta was

estimated at 61 f 8%. This was lower than, but not significantly different from, the 65 î

3% males estimated for Bolinas Lagoon ( z = -0.5, P = 0.6, Fig. 6.1). During winter

(December through February), there were an estimated 65 &9% males in the Fraser

Delta and 62 t 3% males in Bolinas (z = 1 .O, P = 0.3). Sex ratios at both sites were

significantly male-b id ovedl (21 > 33, P < 0.00 1) and in winter (X2i > 34, P < 0.001).

Sex ratio patterns varied throughout the nonbreeding season, but varied similarly for the

two sites (Fig. 6.2). The probability that sex was correctly assigned by our model was

88%.

Body size

After correcting for observer b i s , the mean culmen length of male Duniin in the

Fraser Delta (37.0 mm + 0.1) was estimated to be about 0.2 mm I 0.1 larger (z = 1.3, P =

0.09) than the mean for Bolinas (36.8 mm f 0.1). There was no difference between sites

in female mean culmen lengths (40.6 mm, z = O, P = 0.5).

DISCUSSION

Sex ratios of nonbreeding Dunlin were significantly male-biased in the Fraser

River Delta, both overall (October-April) and during winter (December-February), but

not more so than sites farther south. B u c h a n et al. (1986) found an overall male bias

sirnilar to that of the Fraser Delta (62%) in W i n saniples fiom severai sites in

Washington. Our data also show male biases in the Bolinas Dunlin population overal!

and in winter, findings supported by those of Page (1974). Since males predominate in

163

the Fraser Delta, Bolinas, and in Washington, we must ask whether there is a sex ratio

bias in the entire C. a. pacifica subspecies, and if not, then ask where are the remaining

females during the nonbreeding season?

If there were a sex ratio bias favoring males in the entire C. a. paciJca subspecies,

we would expect to find unmated males on the breedig grounds in sumrner or remaining

on the wintenng grounds during the breeding season, neither of which has been reported

(Paulson 1993, D. Schamel pers. comm.). Altematively, we rnight expect males to have

higher survivorship than femdes, but this is not true for C. a. pacifca (Warnock et al.

1997, P. C. Shepherd, unpubl. data). We corrected for sex differences in habitat use in

the Fraser Delta that would have resulted in a bias toward capturing more females. Al1

else king equal, since smaller males are likely to be more maneuverable than females,

we would expect to catch more females. The opposite is true, so a trapping sex bias dws

not account for our results.

The body-size and dominance hypotheses predict that we should find the

remaining females north of the Fraser Delta. However, too few Dunlin winter north of

the delta for this to account for the difference (Paulson 1993, Warnock and Gill 1996).

These hypotheses also predict that we should find larger individuals in the Fraser Delta.

There was some indication that this may have been tnie of male Dunlin (P = 0.09), but

the statistical results were equivocal and we found no difference in female mean culmen

length between sites. We therefore do not have convincing evidence, either within or

among the sexes, to support the body-size or dominance hypotheses.

The arrival-time hypothesis predicts that we would find the remaining females

south of Bolinas. We examined data h m 25 C. a. pacifca skins fiom the Museum of

Vertebrate Zoology at the University of California at Berkeley and 13 skins fiom the

Natural History Museum of Los Angeles County. We found a 1 : 1 winter sex ratio in

birds collected from a number of sites south of Bolinas over a number of years beginning

in the 1890s. Small samples of Dunlin trapped in San Diego, California (n = 25) and La

Paz, Mexico (n = 5) since 1989 were determined to be male-biased using our mixture

mode1 (B. Kus and R. Carmona, pers. comm.). There are sites south of Bolinas that

support large numbers of Dunlin (Wamock and Giil1996, Page et al. 1999), but the

necessary data on population numbers and proportions of females are currently

unavailable.

Within-latitude habitat segregation between the sexes does not appear to explain

the male bias we found in the Fraser Delta and Bolinas Dunlin populations. Buchanan et

al. (1986) suggested that such segregation might have occurred at two of their study sites

in Washington. They also detected some variation in sex ratios arnong their four main

study sites, although they still found an overall male bias. Radio-telemetry data collected

in the Fraser Delta indicated that habitat segregation between the sexes occuned on a

daily scaie (P. Shepherd, unpubl. data), so we corrected for this in our analysis. Wamock

et al. (1995) also found some habitat segregation between the sexes in Bolinas, but it was

on a seasonal rather than a daily scale. They found that male Dunlin were more likely

than females to make one-way, mid-season movements to agricultural and wetland

habitats up to 140 km inland, and this difference is reflected in the month-to-month sex

ratio results presented here.

We suggest one final possibility. Our results could reflect a latitudinal cline in

sex ratio within two partially overlapping wintering populations, with males wintering

north of females within each population (Fig. 6.3). This would produce what appears to

be a weak cline over the species' range. Twenty-nine of 32 resightings of Dunlin marked

with picric acid in the Yukon-Kuskokwim Delta (the more northerly breeding site)

occurred in Canada and the northwestem United States, while 22 of 24 resightings of

Dunlin marked on the Alaska Peninsula (the more southerly breeding site) occurred in

California (R. Gill, pers. comm.). Thus, our data may reflect a latinidinai cline in sex

ratio within each of two nonbreeding populations, with partial overlap obscuring the

segregation within populations. This hypothesis predicts that we would find female-

biased populations between southem Washington and northern California, and h m

southen California into Mexico. At present, the data requiseci to test this hypothesis are

not available. The Mexico data cited above neither support nor reject the hypothesis,

since the sample sizes are very smail.

The seasonal pattern of sex ratios (Fig. 6.2) is consistent with partiaily

overlapping populations. The percentage of females was higher at both tàe Fraser Delta

and Bolinas in November, perhaps reflecting the movement of females to sites farther

south. Buchanan et al. (1986) found a comparable pattern at the site for wbich they had

the most complete data set (Nisqually, Washington). The proportion of female M i n in

the Fraser Delta hcreased in late winter and early spring, but at the same time the total

nurnbers of birds increased (Shepherd 200 l), suggesting that females moved into the

delta in late winter and early spring. In March, when the number of birds in the Fraser

Delta reached the spring maximum, the proportion of males once again increased. At

Bolinas, the total numbers of Dunlin decreased dong with the percentage of males in late

winter and early spring (Wamock 1994). This could have been due to early northward

movements of males (Paulson 1993), or disproportionate movements of males inland

with the advent of heavy rains (Wamock et ai. 1995).

We conclude that some form of latitudinal cline in sex ratio occurs in C. a.

pacijica. However, additional data h m the Oregon Coast, southem California, and

Mexico are required to determine the location of the missing femaies and to confidently

document the pattern of clinal variation.

ACKNOWLEDGMENTS

We would like to thank al1 of the volunteers in British Columbia and California

who assisted with fieldwork. R. Gill, B. Kus, G. Page, and R. Carmona kindly shared

their unpublished data. The Museum of Vertebrate Zoology at UC Berkeley and the

Natural History Museum of Los Angeles County provided data on the M i n skins in

their collections. R. Elner, F. Cwke, J. Buchanan, D. Dobkin, and an anonymous

reviewer provided insightful comments that impmved the manuscript. Funding for the

project was provided by the Centre for Wildlife Ecology at Simon Fraser University, BC,

the Canadian Wildlife Service, the Point Reyes Bird Obsematory, CA, the James L.

Baillie Memorial Fund of the Long Point Bird Observatory and Bird Studies Canada, and

an NSERC graduate scholarship to P. Shepherd.

FIGURE LECEND

Figure 6.1. Culmen distribution of Dunlin (males, females, and overall population) h m

the Fraser River Delta, British Columbia, estimated using mist-netted and reference

sampies (observed), and h m Bolinas Lagoon, California estimated using the mist-netted

sample (observed) and Bayesian pnor probabilities.

Figure 6.2. Monthly percent male Dunlin (including standard errors and sample sizes) for

the Fraser River Delta, British Columbia and Bolinas Lagoon, California (Fraser River

Delta sample corrected for sex differences in habitat use).

Figure 6.3. Hypothesized patterns of latitudinal clines in sex ratio (more males at higher

latitude ends of each oval) in Dunlin wintering along the Pacific Coast. Pattern 1 (solid

line) may result if there is a latinidinai cline in sex ratio occurring in C. a. pacifica as a

whole, and pattern 2 (dashed line) may result if there are latitudinal clines in sex ratio

occurring in two partially geographically separate populations of C. a. pacifica.

Figure 6.1

140 3 r l Fraser River Delta 1 i Population observed

Population estimated Male estimated Female estimated

Culrnen Length (mm)

8 120 . - Population estimated

. . . . . Male estimated - - Female estimated

I

Bolinas Lagoon O Population observed

Figure 6.2

+ Fraser River Delta, BC

O 1 1 1

Oct Nov Dec Jan Feb Mar Apr

Month

Figure 6.3

Bolinas Lagoon

Single C. a. pacifica nonbreeding population

- . - . - Two partially geographically separate C. a. pacifica non- breeding populations

SIJMMARY, MANAGEMENT IMPLICATIONS, AND GENERAL

IMPLICATIONS

INTRODUCTION

1 investigated aspects of the ecology of non-breeding Dunlin (CaIiuiis alpina

pacijica) in the Fraser River Delta, B. C., the most northerly site supporting a significant

population (approxirnately 40,000 birds) in winter. 1 used radio telemetry, direct

observation, and Geographical Information Systems to document site fidelity, space use

patterns, habitat preferences (at regional and local scaies), and time-activity budgets of

individual Dunlin throughout the 24-hour day and twice-daily tidal cycles. I related

Dunlin behavioural patterns to the densities of their marine invertebrate prey (large

annelids, smail annelids, crustaceans, and molluscs) and to potentiai predation risk. By

following individuals that differed by age, sex, and sub-population (site) and calculating

within-bird means by tide stage, macro-habitat, and time of &y, 1 minimized sampling

biases and produced a detailed picture of the individuals' and the groups' behaviour.

SITE FIDELITY

Summary

Dunlin in the Fraser Delta showed strong within-year site fidelity, both regionally

and locaily, and can be site faithful between years as well, as evidenced by the return of

an individual radio-marked two inters earlier to within 500 m of its original banding

site. Seventy-two percent of 39 radio-marked Dunlin were never absent h m the Fraser

Delta region for more than a day during the study periods. Dunlin did leave the Delta

when ice covered foraging habitat, but their numbers returned to pre-tieeze levels shortly

d e r melt. Locally, there was little home range overlap among Dunlin fiom the eastem,

western, and central parts of the Delta.

Management impücations

Regional fideliîy

Populations that show as much fidelity as do the Dunlin in the Fraser Delta may

be criticaily impacted by the loss of a site, particularly if alternate sites are rare, less

productive, distant, andor already occupied. Fraser Delta Dunlin were able to access

aitemate sites during k z e events, but the fact that they returned to the Delta in large

numbers as soon as the ice melted suggests that the alternate sites were less suitable in

some way.

Local fidelity

Dunlin exhibited high fidelity to sites within the Delta as well as to the Delta

itself. Birds using areas near point sources of contaminants, such as the Iona sewage

outflow, would therefore be expected to be subject to repeat exposure, and could be used

by managers to assess the effects of such exposure. The high level of site fidelity within

the Delta also means that managers could assess Dunlin population trends for the Delta as

a whole by sampling subareas, excluding, of course, areas of potentially high

contaminant input. This would decrease the person-hours required for any single survey

and allow for more frequent sweys, thereby improving the power of trend analyses.

SPACE USE

Summary

Individual non-territorial Dunlin modulated their use of space in close relation to

invertebrate prey density. Across sites, marine home range size decreased as prey density

within the home range increased, with prey density accounting for 63% of the variance in

home range size. Within a single site, both marine home range and core area size

decreased as prey density increased, with prey density explaining 89% of the variance in

home range size and 80% of the variance in core m a size. Dunlin core areas contained

higher densities of crustaceans and small annelids than did their home ranges, indicating

that Dunlin focussed their use of space on the better feetling areas within their ranges.

Mer controlling for the effects of prey density, neither Duniin home range nor core area

size increased with increasing prey patchiness.

Crustacean density alone explained 59% of the variation in home range size

across sites, 60% of the variation in home range within the Mud Bay site, and 73% of the

variation in core area size within the Mud Bay site. However, large annelid density was

similady negatively related to both home range (64%) and core area (75%) size within

Mud Bay, so Dunlin rnay have been cueing in tu some combination of crustacean and

large annelid density at that site. 1 c m o t determine the relative importance to Dunlin of

crustacean and large annelid prey with the data available, but crustaceans appear to be

more generally related (across sites as well as within-site) to the amount of space Dunlin

used. In addition, since my data show conelation rather than causation between

invertebrate density and home range and core area size, 1 cannot be certain that either

crustacean or large annelid density are actual determinants of Dunlin area use.

Management implications

Crustaceans and large annelids may both be important to the maintenance of

Duniin populations in the Fraser Delta. Further research is needed to determine which of

the two, or what combination of the two, is actually king consumed by Duniin, and the

degree of Dunlin diet flexibility. However, for now, data on crustacean density may be

sufficient to predict the amount of marine habitat used by individual Dunlin across the

Delta. Duniin used less space in areas of higher crustacean density. These areas can

support higher densities of Dunlin (see below), and should thecefore be of higher

conservation concern. See "Sub-population effects" below for M e r management

implications.

HABITAT PREFERENCES

Summary

Macmhabitat pmferences (marine venus terrestrial)

Most (> 70%) of the radio-marked Dunlin wintering in the Fraser Delta exhibited

some switching between marine and terrestrial macro-habitats, primarily at night, and

during high tides. Seventy percent is a minimum estimate, since the individuals that were

never located in the terrestrial habitat at night were also twice as likely to be missing

fiom the study area altogether. The radio ranges were considerably shorter in the

terrestrial than in the marine habitat, due to landscape feahires that diminished signal

strength.

The percentage of time Dunlin spent feeding did not differ significantly between

macro-habitats. However, since the tercestrial habitat was rarely used during the day, it

contributed less to the overall energy intake of Dunlin than did the marine habitat. On

average, Dunlin spent 1 1.6 to 14.0 hours per 24-hour day feeding in the marine habitat,

and 4.0 to 2.9 hours feediig in the terrestrial habitat (winter and spring, respectively). In

addition, there was considerable variation among individuals in the use made of the

terrestrial habitat at night. 1 do not have any data on the relative food intake by macro-

habitat, but preliminary fmdings estimate the contribution of the terrestrial macro-habitat

to be approximately 30% of the diet (L. Evans-Ogden, pers comm.). The contribution of

the terrestrial macro-habitat to the diet of individual Dunlin ranged fiom of 1 -92%,

coinciding with tiie telemetry data showing the range of variability in the relative use,

among individuals, made of the terrestrial habitat.

Micro-habitat prderences (witbin the terrestrial zone)

Dunlin chose habitats non-randomly at both regional and local scales. Dunlin

d e d marine habitats highest, but most (> 80% of the Dunlin for which we had a

sufficient sample of locations to calculate a home range) were located in a range of

terrestrial habitats as well. Use of the terrestrial zone occurred primarily at night, when

falcon species, the Dunlins' main predators (see chapter 5), were inactive. Regionally,

soil-based agricultural crops (bare, crop residue, Pasture, and winter cover) ranked above

the remaining twri terrestrial habitats (grassy/unknown and 'other', which included

greenhouses), and Pasture was the only terrestrial habitat that was ranked highly and was

significantly preferred at both scalcs. Locally, Dunlin were not particularly selective of

the other terrestrial habitats within their home ranges.

Pastures in the Fraser Delta are heavily fertilized with cattle manure, and support

abundant terrestrial invertebrates (Fratello et al. 1989). Pasture vegetation tends to be

short, and densities of Duniin using pastures in California correlateci negatively with

vegetation height (Colwell and Dodd 1995). Wet bare fields may attract Dunlin because

they resemble mudflats, as do some of the crop residue fields (but with a little additional

vegetation). Winter cover m p s may be less attractive io Dunlin because they can grow

quite tall, obscuring the view of approaching predators, however, they rnay be used once

the waterfowl wintering in the area have grazed back the vegetation (Taitt 1997).

Management implications

The importance of the terrestrial macro-habitat to Dunlin rnay have previously

k e n underestimated, since that habitat was used far more at night than during the day.

Alternatively, Duniin rnay be using the terrestriai habitat to a greater extent now that they

did in the pst. In any case, to date, avian management plans for the terrestrial habitats

have focused on the needs of raptors, waterfowl, and passerines, birds whose

requirements rnay differ fiom those of Duniin. For example, raptors perch and forage in

wildland habitats and areas covered in tall grasses; waterfowl may a h choose habitats

with taller vegetation than would be preferred by Dunlin; hedgerows have been promoted

as habitat for passerines, but fiagmentation of fields and the addition of perching and

hiding sites for raptors rnay be detrimental to Dunlin.

Access to terrestrial habitats as an alternative food source within close proximiiy

to the intertidal zone rnay be required for some Dunlin to meet their energy needs

(Velasquez and Hockey 199 1, Davidson and Evans 1986, Rottentmm 1996, Weber and

Haig 1996, Dam 1999). The terrestrial habitat may allow for increased food intake

during high tides when marine habitat is less available or altogether submergecl, or during

periods of rainfall when fieshwater input rnay decrease the availability of marine

invertebrates, particularly for shorter-billed birds, while increasing the availability of

terrestrial invertebrates (Goss-Custard 1970, Gerstenberg 1979, Goss-Custard 1984,

Pienkowski 198 1, Heitmeyer et al. 1989).

Further research focussing on the terrestrial zone rnay clarifi how much the

terrestrial habitat contributes to the Dunlin population's energy budget, and the relative

importance of each of the terrestrial micro-habitats (L. Evans-Ogden, pers. comm.). In

the meantirne, maintaining as much soil-based agricdture, with an emphasis on pasture,

on the land near intertidal mudflats, would be preferred for Dunlin in the Fraser Delta and

other wintering areas near to agricultural lands. Fragmentation should be kept to a

minimum, since Dunlin preferred large fields, likely in response to predation risk.

The fact that Dunlin forage in terrestriai as well as marine habitats should be

considered when determinhg whether Dunlin would be suitable indicators of the levels

of toxins in the estuary. In the case of agricultural pesticides, the toxins codd

accumulate in W i n tissues via agricultural rwisff into the estuary, or h m direct

consumption by Dunlin foraging in terrestrial habitat.

TIME-ACTMTY BUDGETS

Summary

Dunlin in the Fraser Delta, at the northem end of the non-breeding distribution in

which they are common, are subject to high themoregulatoty costs and they spent on

average 3 hours per day in flight. To meet these high energetic costs, Dunlin spent more

than 60% of their tirne foraging, both day and night, and, for the rnost part, in both

marine and terrestrial habitats. Dunlin spent on average at least 15.7 hours per 24-hour

day foraging. At leaset 7.1 of those hours occurred at night, and at least 2.9 of the night-

time feeding hours occurred in the terrestrial macro-habitat. Mer feeding and flying,

Dunlin had at most of 5.3 hous per 24-hour day for al1 other activities, such as rwsting,

preening, vigilance, and aggression. Dunlin spent more time feeding in 1996 than in

1998, and in spring than in winter, both during the day and at night.

Management implications

Dunlin have at most 5.3 hours to spare during the 24-hour day ( d e r foraging and

flying), so any serious negative impact on their foraging habitats is likely to result in a

decline in the population. Dunlin foraged prirnarily in marine habitat, however, those

Dunlin that did use the terresirial habitat (> 70% of the radio-marked individuals), used it

prirnarily for foraging (more than 60% of the time spent there). Although not al1 Dunlin

used the terresnial habitat, early indications h m L. Evans-Ogden's research estimates

that approximately 30% of the Fraser Delta population's diet cornes fiom the terrestriai

habitat, as mentioned above, and some individuals obtain as much as 92% of their food

there (L. Evans-Ogden, pers cornm.).

Dunlin in the Fraser Delta spent a significaut amount of time and energy in flight,

likely as a rneans of decreasing predation risk (Brennan et al. 1985, Creswell 1994,

Dekker 1998 and 1999, Hotker 2000). Peregrines hunting Dunlin here were successfiil

on 33% of surprise attacks and only 8% of aerial chases (Dekker 1998 and 1999). Should

the numbers of falcons wintering in the Delta continue to increase, not only might Duniin

mortality tates increase, but Dunlin might also reach a point where they are unable to

balance their energy budgets. The numbers of Dwilin in the Fraser Delta might therefore

decline, either by increased rates of mortality or by the birds moving to other sites.

AGE EFFECTS

Summary

Home range size did not differ between Dunlin of different age classes, nor were

there any differences in the densities of the four invertebrate types within adult and

juvenile home ranges. The relative use made of the terrestrial habitat did not differ

significantly between age classes, nor was there any significant difference in the percent

of time spent foraging. As a caveat, 1 would like to add that 1 could only confidently age

the Dunlin trapped early in the season (winter 1995/96), so my power to test for age

effets was limited.

SEXEFFECTS

Summary

There were behavioural differences between the sexes. Femaie Dunlin had

significantly larger marine home ranges than males. This difference between the sexes

did not appear to be due simply to body size differences, since PCl was not related to

marine home range or core area size overall (across sites). The difference between the

sexes in Dunlin home range size may have been related to differences in the density of

food within their respective home ranges, however the evidence to support this is

tenuous. Tbere was some indication that crustaceans, which are important prey items for

pcifca Duniin, rnay have been l e s dense within femaie home ranges than those of

males (P = 0.06, Table 2.9). Crustaceans were significantiy less dense within female than

male core areas (P = 0.04, Table 2.9). In addition, there were lower densities of al1 four

invertebrate types (although not statistically significantly individually) within the home

ranges of females than those of maies.

Maies were more likely to be located in the terrestriai habitat (64.4 % + 5.8 of

locations) than females (43.3 % 2 7.3) (FlJ5 = 5.1, P = 0.03, see chapter 3). This finding

is consistent with the hypothesis that shorter-bilted birds (primarily males) had

diminished access to sub-surface marine invertebrates and were therefore more likely to

switch foraging habitats. Male and female Dunlin also exhibited some differences in

their micro-habitat preferences. Females were more selective, exhibiting more

statistically sigiificant preferences in their micro-habitat rankings.

Femaies were larger (PC 1) than males, but they spent less time foraging.

Assuming that both sexes were either meeting their energy requirements, or losing and

gaining weight at the same rate (Kaiser and Gillingham 1985), females must either have

k e n consurning more or better quality food per unit time than males; females did not

conserve energy by spending less t h e in flight. The fact that the marine home ranges of

females were approximately 20% larger than those of males, and may contain lower

densities of crustaceans, does not appear to put femdes at a performance disadvantage in

terms of food intake rate compared to males.

Management implications

Male Dunlin in the Fraser Deita spent more time feeding than females, so if food

availability were to decline (necessitating an increase in feeding time), males may run out

of time before femaies (mail there are at most only 5.3 hours per day during which

Dunlin are not already feeding or flying). There was some indication that males may also

be more likely than females to use the terrestrial macro-habitat as an alternative foraging

site. The aiteration or removai of the terrestrial macro-habitat may therefore put males at

a disproportionate disadvantage. Female Duniii, which spend l e s time in the terrestrial

macro-habitat than maies, may be more usefui than males for monitoring toxins (such as

heavy metals) that are thought to enter the estuary via river runsff.

SEX RATIOS

Summary

Sex ratios of nonbreeding Dunlin were significantly male-biased in the Fraser

River Delta, both overall (October-April, 61% males) and during winter (December-

Febniary, 65% males), but not more so than in Bolinas Lagoon (overall65% males, and

winter 62% males). Monthly sex ratios were similar at the two sites but varied

throughout the winter, presumably reflecting differences in seasonal movement patterns

between the sexes. Since there is no evidence to support the possibility of a skew toward

males in C. a. pacifica as a whole, our data are consistent with some form of latitudinal

cline in the sex ratio of C. a. pacifica. However, additional data from the Oregon Coast,

southem California, and Mexico are required to determine the location of the "missing"

females and to confidently document the pattern of clinal variation. We also tested the

hypothesis that mean body size within each sex is larger at the higher-latitude site (Fraser

River Delta), where ecological variables might favour larger birds, but this did not appear

to be the case.

Management implications

In species such as Dunlin that appear to show a latitudinal cline in sex ratio, a

catasûophic event like an oil spi11 covering a portion of the non-breeding range may

affect one sex disproportionately.

SüB-POPULATION EFFECTS

Summary

Dunlin were highly site-faithful within the Fraser Delta; there was little home

range overlap among Duniin h m the eastern, western, and central parts of the Delta.

Bir& trapped in areas of sandier sediment were l e s likely to be located in areas of

muddier sediment and vice versa.

Dunlin home range and core m a sizes differed among the three sites, and were

largest at the site where ovedl invertebrate prey density within the spaces used was

lowest, and where sediment grain size was largest (Boundary Bay). In addition, Dunlin

density was lowest at Boundary Bay (102 birds/km2), and was highest at Westham Island

(192 birds/km2), when prey dnisity was among the highest. In spite of differences in

marine prey density, Dunlin density, and marine home cange and core area sizes, 1 found

no differences in indices of Dunlin performance (body size, M y weight, and percent

time spent fding) among sites within the Fra= River Delta.

W i n exhibited different micro-habitat preferences among sites, at both regional

and local d e s . There was considerable variation in the site-specific ranking of

terrestrial habitats, and many of the habitats that ranked high were not statistically

significantly preferred over those ranlced low. There was some indication that Dunlin

h m Boundary Bay may have been more likely than birds from the other two sites to be

located in the terrestrial macro-habitat (P = 0.06).

Management implications

Vertebrate species tend to use less space and congregate in higher densities where

habitat is more productive (Harestad and Bunnel 1979, Yates et al. 1993). Dunlin density

was lowest and the arnount of space they used was largest where oved l invenebrate prey

density within the birds' ranges was lowest (Boundary Bay). This is not to say that

Boundary Bay was not important to Dunlin. Birds fiom this site used more space in

order to obtain the food energy they required, but they did not appear to be at an overail

perfonance disadvantage compared to Dunlin h m areas of higher prey densities.

Should the Boundary Bay area be removed from the habitat base, those Dunlin would

probably move into the smunding mas, thereby increasing interference among al1

individu&, and possibly decreasing mean food intake rate throughout the population

(Stillman et al. 1997 and 2000). However, if one had to list the sites by their relative

importance to Dunlin, based on W i n population densities and the densities of their prey

populations, Roberts' Bank north of the coal port jetty should be the highest pnority for

conservation, followed by eastem Boundary Bay (Mud Bay), and western Boundary Bay.

Dunlin density was lowest where mean intertidal sediment grain size was largest

where (in Boundary Bay and south of the coal port jetty), and Dunlin density was highest

where sediment grain size was smaiiest (Westham Island). If cause and efiect can be

established, this bas implications for the effects on M i n and other shorebird species of

water flow and sediment deposition changes in the Delta. Before the construction of the

coal port jetty and the Tsawassen ferry jetty, sediment and organic material flowing down

the Fraser River was deposited across the entire intertidal area from Westham Island

south to Pt. Roberts (Figure 2.1). At present, the area south of the coal port jetty no

longer receives much of this input. Finer sediment has washed away, increasing the

mean sediment grain size. There has also been considerable encroachment by a large-

leafed and robust eelgrass species (Zostera marina), which is avoided by foraging

Western Sandpipers (P. Shepherd, T. Sutherland, and B. Elna, unpublished data). Any

changes to water flow into the intertidal zone of the Fraser Delta region should be

carefully considered, and the effects on mean sediment grain size assessed. Changes that

may increase mean sediment grain size may negatively impact the usehlness of an area

to shorebirds.

DunIin showed considerable variation among sites in their ranking of terrestrial

micro-habitats. Overall, Dunlin preferred soil-based agricultural crops, but there was

variation in ranking among sites, and consecutive ranks oAen did not differ statistically

from one another. Maintainhg a mosaic of soil-based agricultural crops throughout the

study area should therefore meet the needs of Dunlin fiom al1 hree sites. Access to

agricultural habitats may be more important to Dunlin h m the Boundary Bay site

(which may use it to a greater extent than Dunlin fiom the other two sites, P = 0.06, Table

2.4), since marine prey density was lowest at this site.

GENERAL IMPLICATIONS

The spatial relationship between home range size and habitat pductivity has

rarely been examineci among individuals within a species. In the Fraser River Delta, 1

determined that variation in marine prey density has substantiai power to account for the

observeci variation in Dunlin home range size, with larger home ranges occuccing in areas

of lower prey density. in addition, Dunlin focussed their use of space in areas within

their home ranges where prey density was above the cange average. Considering that this

is a non-territorial, fl ocking species of bird, the closeness of this relationship is

surprising. In contrast with territorial species, non-territorial species are not subject to a

substantial incteasing disincentive (defense costs) as the amount of space they use

increases. Birds dso have an ease of movement not experienced by flightless vertebrates.

M i n in the Fraser River Delta could fly the length of the Delta in a matter of minutes.

In contrast to ditary species, individuals of flocking species may, in orâer to maintain

group membership, make movements initiated by other flock members and over which

they rnay have little or no trajectory control. The amount of space used by a flocking

individual rnay be influenced in part by other members of the flock, however, we still see

a close relatiomhip among individuals between food density and space use.

The residual percentage of time Dunlin spent foraging, an index of performance,

did not Vary with prey density within the home range. Mer accounting for the time

spent foraging and flying, îhese Dunlin, wintering at the norihem end of their non-

breeding distribution, had at most 5.3 hours per 24-hou day lefi over for other activities.

Time rnay therefore be a lirniting factor for Dunlin in the Fraser Delta, and for other

animals wintering at temperate sites near the edge of their ranges, which could help to

explain the close relationship observed between food density and space use. If the time

available for foraging i s a limiting factor, and if food is plentiful and dependably

available, individuals rnay benefit fiom rninimizing the time speni exploring for food.

Dunlin may also benefit from minirnizing the time spent feding in order to increase the

tirne available for vigilance.

Temperature is often cited as a factor limiting the distribution of species. Small

nwnbers of Dunlin winter as far north as Alaska, so temperature may not be the sole

explanation for the fact that so few Dunlin winter north of the Fraser River Delta. Rather,

significant wintering populations of Duniin rnay be excluded from sites north of the Delta

due to the interactions among temperature, proximity of suitable alternative sites,

disturhce by predators (and pertiaps humans at some sites), and habitat availability. In

the event of a k z e , which c m last for weeks in the Fraser Delta, Dunlin require

sufficient reçentes to either fast until melt or to fly (presumably south) to an alternative

site, as most birds did during this study. Dunlin rnay be able to acquire sufficient fat

reserves to fuel a longdistauce flight to an alternative site (in the absence of predation

disturbance), since food is plentiful. However, Dunlin are subject to daily disnubance by

ptedators, and the level of their fat reserves changes hmughout the non-breeding season

in a manner consistent with the hypothesis that they are modulating theu reserves to

balance the trade-off between starvation nsk and predation risk. Predation risk in the

Delta may thecefore liait the amount of fat individuals can 'safely' carry, thereby

limiting the distance Dunlin could fiy to an alternative site suitable to support a large

Mux of birds.

Advances in satellite telemetry tmhnology may swn make it possible to ttack

small migratory species over greater distances and longer periods of time. Resemhers

may be able to follow known individuals over entire yearly cycles; from the original

breeding site, through migration, to non-breeding sites, and back again. We could use

this technology to describe migration routes and chronology, between-year breeding and

wintering site fidelity, and connections, if they exist, between particular breeding and

wintenng sites. Radio telernetry and Geographical Information Systems cm hopefully be

used, more and more, to guide both wildlife managers and theoretical ecologists, to better

inform sustainable wildlife conservation strategies, and to extend the theoretical

parameters of iùture research.

Aebischer, N. J., P. A. Robertson, and R. E. Kenward. 1993. Compositional andysis of habitat use fiom animal radio-tracking data. Ecoiogy 74: 1 3 13-1 325.

Ashicenazie, S., and U.N. Safriel. 1979. Breedig cycle and behavior of the Semipalrnated Sandpiper at Barrow, Alaska. Auk 9656-67.

Atkinson, N. K., R. W. Surnmers, M. NicolI, and J. J. D. Greenwood. 1981. Population, movements and biometrics of the Purple Sandpiper Calidris maritima in eastem Scotland. Omis Scand. 12: 18-27.

Bailey, R. C., and J. Byrnes. 1990. A new, old method for assessing measurement error in both univariate and multivariate morphometric studies. Syst. Zool. 39: 124-130.

Baker, A. J., T. Piersma, and A. D. Greendade. 1999. Molecular vs. phenotypic sexing in Red Knots. Condor 101 :887-893.

Baldwin, J. R. and J. R. Loworn. 1994. Habitats and tidal accessibility of the marine foods of dabbling ducks and brant in Boundary Bay, British Columbia. Mar. Biol. 12O:627-638

Barrett, R. T., M. Peterz, R. W. Furness, and J. Durinck. 1989. The variability of biometric measurements. Ringing and Migr. 10: 1 3-1 6.

Belthoff, J. R., and S. A. Gauthreaux Jr. 1991. Partial migration and differential winter distribution of House Finches in the eastern United States. Condor 93:374-382.

Beyer, D. E. Jr., and J. B. Haufler. 1994. Diurnal versus 24-hour sampling of habitat use. J. Wildl. Manage. 58: 178-1 80.

Bildstein, K.L., G.T. Bancroft, P.J. Dugan, D.H. Gordon, K.M. Erwin, E. Nol, L.X. Payne, and S.E. Semer. 1991. Approaches to the conservation of wetlands in the western hemisphere. Wilson Bull. lO3:2 1 8-254.

Boates, J.S. and P.C. Smith. 1989. Crawling behaviour of the amphipod Corophium volutator and foraging by Semipalrnated Sandpipers, Calidris pusilla. CM J. Zool. 67:457-462.

Brennan, L. A., J. B. Buchanan, C, T. Schick, S. G Herman, and T. M. Johnson. 1984. Sex determination of Dunlins in winter plumage. J. Field Omithol. 55343-38.

Brennan, L.A., J.B. Buchanan, S.G. Herman, and T.M. Johnson. 1985. interhabitat movements of wintering Duniin in western Washington. Murrelet 66: 1 1 - 16.

Brennan, L.A., M. A. Finger, J. B. Buchanan, C. T. Schick, and S.G. Hennan. 1990. Stomach contents of Dunlins collected in western Washington. Northwest. Nat. 71:99- 102.

Brown, J. L. 1975. The evolution of behaviow. Norton Press, NY.

Brown, S., C. Hickey, and B. Harrington. 2000. The U. S. Shorebird Conservation Plan. Manomet Ctr. for Cons. Sci. Manomet, MA.

Browning, M. R. 1977. Geographic variation in Ddins, Calidris alpina, of North America. Cm. Field-Nat. 91:391-393.

Bryant, D. M. 1979. Effects of prey density and site-character on estwry usage by overwintering waders (Charadrii). Est. and Coastal Mar. Sci. 9:369-3 84.

Buchanan, J.B., L.A. Brennan, C.T. Schick, M.A. Finger, T.M. Johnson, and S.G. He-. 1985. M i n weight changes in relation to food habits and available prey, J. Field ûmithol. 56:265-272.

Buchanan, J. B., L. A. Brennan, C. T. Schick, S.G. Herman, and T. M. Johnson. 1986. Age and sex composition of wintering Dunlin populations in western Washington. Wader Study Group Bull. 46:374l.

Buchanan, J.B., C.T. Schick, L.A. Brennan, adn S.G. Herman. 1988. Merlin predation on wintering Dunlins: hunting success and Dunlin escape tactics. Wilson Bull. 100: 108-1 18.

Burger, J. 1986. The effect of human activity on shorebirds in two coastal bays in norheastern United States. Environ. Cons. t 3: 123-130.

Burger, J. 1988. Effects of demolition and k h clean-up operations on birds on a coastai muâfiat in New Jersey. Est., Coastal and Sheif Sci. 27:95-108.

Burton, N. H. K. 2000. Winter site-fidelity and survival of Redshank Tringa totanus at Cardiff, south Wales. Bird Study 47: 102- 1 1 2.

Burton, N. H. K. and P. R. Evans. 1997. Survivai and winter site-fidelity of Tumstones Arenaria interpres and Purple Sandpipers Caiidris maritirna in northeast England. Bud Study 44:35-44.

Butler, R. W. 1992. Abundance, distribution and conservation of birds in the vicinity of Boundary Bay, British Columbia. Cm. Wildl. Sem. Tech. Rep. No. 155, Ottawa.

Butler. R. W. 1999. Winter abundance ad distribution of shorebirds and songbirds on f d a u d s on the Fraser River delta, British Columbia, 1989-1991. Can. Field Nat. 1 13:390-395.

Butler, R.W. and R.W. Campbell. 1987. The birds of the Fraser River delta: populations, ecology and international significance. Can. Wildl. S ~ N . Occas. Paper No. 65. Ottawa.

Butler, R W., and K. Vermeer. 1994. The abundance and distribution of estuarine b i s in the Strait of Georgia, British Columbia. Can. Wildl. Sew. Tech. Rep. No. 83.

Caldow, R. W. G., J. D. Goss-Custard, R. A. Stillman, S. E. A. Le V. dit Durell, R. Swinfen, and T. Bregnballe. 1999. Individual variation in the cornpetitive ability of interference-prone foragers: the relative importance of foraging efficiency and susceptibility to interference. J. Anim. Ecol. 68:869-878.

Calder, W. A. 1974. The consequences of body size for avian energetics, p. 86- 15 1. in R. A. Paynter [ed.]. Avian energetics. Nuttali Ornithological Club 15, Cambridge, MA.

Campbell, R. W., D. A. Manuwal, and A. S. Harestad. 1987. Food habits of the Common Barn-Owl in British Columbia. Can. J. Zwl. 65578-586.

Casiro, G. and J. P. Myers. 1989. Flight range estimates for shorebirds. Auk 106:474-476.

Clark, K. E., L. J. Niles, and J. Burger. 1993. Abundance and distribution of migrant shorebirds in Delaware Bay. Condor 95:694-705.

Colwell, M.A. 1993. Shorebird community patterns in a seasondy dynarnic estuary. Condor 95: 104-1 15.

Colwell, M.A. and S.L. Landrum. 1993. Nonrandom shorebird distribution of fine-scale variation in prey abundance. Condor 9594- 103.

Colwell, M. A. and S. L. Dodd. 1995. Waterbird communities and habitat reiationstiips in coastal pastures of northern California. Cons. Biol. 9:827-834.

Colwell, M.A. and S.L. Dodd. 1997. Environmental and habitat correlates of pasture use by nonbreeding shorebirds. Condor 99:337-344.

Connors, P.G., J.P. Myers, C.S.W. Connors and F.A. Pitelka. 1981. Interhabitat movements by Sanderlings in relation to f o e n g profitability and the tidal cycle. Auk 98:49-64.

Cresswell, W. 1994. Age-dependent choice of redshank (Tringa rotanus) feediig bcation: profitability or risk? 3. Anim. Ecol. 63589-600.

Cresswell, W. 1996. Surprise as a winter hunting strategy in sparrowhawks Accipiter nisus, peregrines Falco peregrinus and merlins F. coltimbarius. Ibis 138: 684-692.

Cuadrado, M., J. C. Senar, and J. L. Copete. 1995. Lo ail Blackcaps Sylvia atricapilla show winter site fidelity? Ibis 13770-75.

Dann, P. 1999. Feeding periods and supratidal feeding of Red-necked Stints and Curlew Sandpipers in Western Port, Victoria Emu 99~218-222.

Dann, S., D. Masman, A. Sûijkstra, and S. Verhulst. 1989. tntraspecific ailornetry of basal metabolic rate: relations with body size, temperature, composition and circadian phase in the Kestrel, Falco tinnunculus. J. Biol. Rhythms 4:267-283.

Davidson, N.C. and P.R. Evans. 1986. The role and potentiai of man-rnade and man- modified wetlands in the enhancement of the swival of overwintering shorebirds. Colonial Waterbirds 9: 176- 188.

Dawe, N. K., C. S. Runyan, and R. McKelvey. 1978. Seasonal food habits of the Barn Owl (Tyto alba) on the Alaksen National Wildlife Area, British Columbia. Can. Field- Nat. 92:151-155.

Dekker, D. 1998.Over-ocean flocking by Dunlins, Calidris alpina, and the effect of raptor predation at Boundary Bay, British Columbia. Can. Field-Nat. 1 12:694-697.

Dekker, D. 1999. Bolt from the Btue: Wild Peregrines on the Hunt. Hancock House Publ. Ltd., Surrey, B. C.

Dierschke, V. 1998. High profit at high risk for juvenile Dunlins Calidris alpina stopping over at Helgoland (German Bight). Ardea 86: 59-69.

Dodd, S. L. and M. A. Colwell. 1996. Seasonal variation in diumal and noctumal distributions of nonbreeding shorebirds at North Humboldt Bay, California. Condor 98: 196-207.

Dodd, S. L. and M. A. Colwell. 1998. Environmental correlates of diumal and nocturnal foraging patterns of nonbreeâing shorebirds. Wilson Bull. 1 10: 182-189.

Donaidson, G, C. Hyslop, R1.G. Momson, 1. Davidson. 2001(in press). Canadian Shorebird Conservation Plan. Can. Wildl. Serv., Env. Canada, Hull, QC. 32pp.

Dugan, P.J. 1981. The importance of noctumal foraging in shorebirds: a comequence of increased invertebrate prey activity. in: N.V. Jones and WJ. Wolff [eds.]. Feeding and survival strategies of esturine organisms. Plenum Press, N.Y.

Dunk, J. R. and R. J. Cooper. 1994. Territory-size regulation in Black-shouldered Kites. Auk 11 1 :588-595.

Dunn, M., D. McCullough, C. Kernbie, S. Freeman, and J. Komaromi. 1993. Southern Roberts Bank and Boundary Bay, B. C. vegetative communities. Environ. Canada Rep.

Engelmoer, M., and C. S. Roselaar. 1998. Geographicai variation in waders. Kluwer Academic Publishers, Dordrecht, Netherlands.

Ens, B. and J.D. Goss-Custard. 1984. Interference arnong oystercatchers Haematopus osûaiegus, feeding on mussels, Mytilus edulis, on the Exe estuary. J. Anim. Ecol. 53:217- 23 1.

Evans, A.D. 1987. Relative availability of the prey of wading birds by day and by night. Mar. Ecul. Progr. Ser. 37: 103-1 07.

Evans, P.R. 1976. Energy balance and optimal foraging strategies in shorebirds: some implications for their distribution and movements in the non-breeding season. Ardea 64: 1 17-1 39.

Evans, P.R. 198 1. Migration and dispersal of shorebirds as a sumival strategy. In: N.V. Jones and W.J. Wolff [eds.] Feeding and survival strategies of estuarine organisms. Plenum Press, London.

Evans, P. R. 1986. Experimental evidence for the use of visual cues by foraging Dunlins. Wader Study Group Bull. 48: 14-15.

Evans, P.R. 199 1. Seasod and annual patterns of moctality in migratory shorebirds: some conservation implications, pp. 346-359. In: C.H. Pemns, J.-D. Lebreton, and G.J.M. Herons [eds.]. Bird Population Studies. Oxford University Press, Oxford.

Exo, K.-M., U. Eggers, R Laschefski-Sievers, and G. Scheiffarth. 1992. Monitoring activity patterns using a micro-computerîontrolled radiotelemetry system, tested for waders (Charadrii) as an example, pp. 79-87. In: 1. G. Pnede and S. M Swift [eds.]. WiIdlife telemeîry, remote monitoring and tracking of animals. Ellis Horwood, Chichester.

Exo, K.-M. 1993. Methods to monitor time and energy budgets of birds, especiaily waders (Charadriiformes). ii Int. Symp. Biotelem.:224-23 1.

Fisheries and Environment Canada. 1977. Surficial materials of the southwestern Fraser towland. Fraser River Delta and estuary region map senes. Lands directorate, Fisheries and Environ. Canada.

Fratello, B., M. A. Sabatini, L. Mola, C. Uscidda and C. Gessa. 1989. Effects of Agricultud Practices on Soil Arthropoda: Organic and Mineral Fertilizers in Alfalfa Fields. Agi. , Ecosyst. and Environ. 27:227-23 9.

Freed, L. A., T. B. Smith, J. H. Carothers, and 3. K. Lepson. 1996. Shrinkage is not the most likely cause of bill change in liwi: a rejoinder to Winker. Cons. Biol. 10:659-660.

Fretwell, S. D. 1972. Populations in a seasonal environment. Princeton University Press, Princeton, NJ.

Fretwell, S. D. and H, L. Lucas. 1970. On territorial behavior and othet factors influencing habitat distribution in birds. 1. Theoretical development. Acta Biotheoretica X E : 1 6-36.

Gass. L. 1979. Territory tegulation, tenue, and migration in d o u s hummingbirds. Cm. J. Zool.57:9 14-923.

Gass, C. L., G. Angehr, and J. Centa. 1976. Regulation of food supply by feeding tenitoriality in the nifous hummingbird. Can. J. Zool. 54:2046-2054.

Gerritsen, A.F.C. and Y.M. van Heezik. 1985. Subsirate preference and substrate related foraging behaviour in three Caldris species. Neth. J. Zool. 35:671-692.

Gerstenberg, R. H. 1979. Habitat utilization by winteting and migrating shorebùds on Humboldt Bay, California. Stud. Avian Biol. 2:33-40.

Gibb, J. 1954. Feeding ecology of tits, with notes of tree-creeper and goldcrest. Ibis 96:s 1 3-543.

Gill, F. B. and L. L. Wolf. 1975. Economics of feeding territoriality in the Golden- winged Sunbird. Ecology 56:333-345.

Gill, J. A. and W. J. Sutherland. 2000. Predicting the consequences of human disturbance h m behavioural decisions, pp. 51-64. In: L. J. Gosling and W. J. Suthedand [eds.]. Behaviour and Conservation. Cambridge University Press, Cambridge.

Gjerde, 1. And P. Wegge. 1987. Activity patterns of Capercaillie, Tetrao urogalh, during winter. Holarctic Ecology 10:286-293.

Goede, A.A. and M. De Bruin. 1985a. Selenium in a shorebud, the Dunlin, fiom the Dutch Waddenzee. Mar. Poliution Bull. 16: 1 15- 1 17.

Goede, A.A. and M. De Bruin. 1985b. Arsenic in the Dunlin (Calidris alpina) h m the Dutch Waddenzee. Bull. of Env. Contamin. and Toxicol. 34:617-622.

Gosler, A.G., J.J.D. Greenwood, and C.Perrins. 1995. Predation risk and the cost of king fat. Nature 377:621-623.

Goss-Custard, J.D. 1969. The winter feeding ecology of the Redshank. Ibis 11 1338-356.

Goss-Custard, J.D. 1970a. The responses of redshanks (Tringa totanus)(L.) to spatial variations in the density of their prey. J. Anim. Ecol. 39:91-113.

Goss-Custard, J. D. 1970b. Feeding dispersion in some overwintering wading birds, pp. 3-35. In: J. H. Crook [ed.]. Social Behaviour in Birds and mammals. Academic Press, NY.

Goss-Custard, J.D. 1970c. Factors affecting the diet and feeding rate of the Redshank, pp. 101-1 10. Animal populations in relation to their food resources. Bit. Ecol. Soc. Symp. No. 10, Blackwell, Oxford,.

Goss-Custard, J. D. 1976. Variation in the dispersion of Redshank Tringa totanus on their winter feeding grounds. Ibis 118:257-263.

Goss-Custard, J.D. 1977a. Predatory responses and prey mortality in Redshank, Tringa totanus (L.), and a preferred prey, Corophium volutator (Pallas). J. Anim. Ecol. 46:21-35.

Goss-Custard, J.D. 1977b. The energetics of prey selection in Redshank, Tringa totanus (L.) in relation to prey density. J. Anim. Ecol. 46: 1 - 19.

Goss-Custard, J.D. 1977c. The ecology of the Wash. III. Density-related behaviour and the possible effects of a loss of feeding grounds on wading birds (Charadrii). J. Appl. Ecol. 14:721-739.

Goss-Custard, J.D. 1979. Effect of habitat l o s on the numbers of overwintering shorebirds. Stud. Avian Biol. 2: 167- 177.

Goss-Custard, J.D. 198 1. Feeding behaviour of Redshank, Tringa totanus, and optimal foraging theory, pp. 115-134. In: A.C. Kamil and T.D. Sargent [eds.]. Foraging Behaviour: Ecological, Ethological, and Psychological Approaches. Garland STPM Press, New York..

Goss-Custard, J.D. 1984. Intake rates and food supply in migrating and wintering shorebuds, pp. 233-270. In: J. Burger and B.L. Olla [eds.]. Shorebirds: Migration and Foraging Behaviour. Plenum Press, New York.

Goss-Custard, J.D., D.G. Kay and R.M. Blundell. 1977. The density of migratory and overwintering redshank, Tringa totanus L. and curlew Numenius arquata L. in relation to the density of their prey in southeast England. Estuar. Coast. Mar. Sci. 5:497-5 10.

Goss-Custard, J. D., S. E. A. Le V. Dit Durell, S. McGrorty, and C. J. Reading. 1982. Use of mussel Mytilus edulis beds by Oystercatchers Haematopus ostraiegus according to age and population size. J. Anim. Ecol. 5 1 543454.

Goss-Custard, J.D. and S.E.A. Le V. dit Durell. 1987. Age-related effects in oystercatchers Haematopus ostralegus, feeding on mussels Mytilus edulis. 1. Foraging eaciency and interference. J. Anim. Ecol. 56:52 1-536.

Goss-Custard, J.D. and S.E.A. Le V. dit Durell. 1987. Age-related effects in oystercatchers Haematopw ostralegus, feeding on mussels Myrilus edulis. III. The effect of interference on overall intake rate. J. Anim. Ecol. 56549-558.

Goss-CustaK1, J.D. and M.E. Moser. 1988. Rates of change in the nurnbers of Dunlin, Calidris alpina, wintering in British estuaries in relation to the spread of Spartinu anglica. J. Appl. Ecol. 2S:W- 1 O%

Goss-Custard, J.D. and S.E.A. Le V. dit Durell. 1990. Bird behaviour and environmental planning: approaches in the study of wader populations. Ibis 132:273-289.

Goss-Custard, J.D., R.M. Warwick, R Kirby, S. McGrorty, R.T. Clarke, B. Pearson, W.E. Rispin, S.E.A. Le V. Dit Durrell, and R.J. Rose. 1991. Towards predicting wading bird densities from predicted prey densities in a pst-barrage Severn estuary. J. Appl. Ecol. 28: 1004-1026.

Goss-Custard, J.D., R.W.G. Callow and RT. Clarke. 1992. Correlates of the density of foraging oystercatchers Haematopus osrralegus at different population sizes. J. Anim. Ecol. 61:159-173.

Goss-Custard, I. D., R. T. Clarke, K. B. Briggs, B. J. Ens, K.-M. Exo, C. Smit, A. J. Beintema, R. W. G. Caldow, D. C. Catt, N. A. Clark, S. E. A. Le V. Dit Durell, M. P. Harris, J. B. Hdscher, P. L. Meininger, N. Picozzi, R. Prys-Jones, U. N. Sahiel, and A. D. West. 1995a. Population consequeues of winter habitat loss in a migratory shorebird. 1. Estimating mode1 parameters. J. Appl. Ecol. 32:320-336.

Goss-Custard, J. D., R. T. Clarke, S. E. A. Le V. Dit Durell, R. W. G. Caldow, and B. J. Ens. 1995b. Population consequences of winter habitat loss in a migratory shorebird. 1. Mode1 predictions. J. Appl. Ecol. 32:337-351.

Goss-Custard, John .D., John Ross, Selwyn McGrorty, Sarah E. A. Le V. Dit Durrell, Richard W. Caldow, and An- D. West. 1998. Locally stable wintering numbers in the Oystemtcher Huematopras osfralegus where carrying capacity bas not been reached. Ibis 140:104-112.

Goss-Custatd, J. D., R. A. Stillrnan, A. D. West, S. McGrorty, S. E. A. Le V. Dit Durell, and R. W. G. Cddow. 2000. The d e of behaviowai models in predicting the ecological impact of harvesting, pp. 65-82. in: L. J. Gosling and W. J. Sutherland [eds.]. Bebaviour and Conservation. Cambridge University Press, Cambridge.

Goudie, R. 1. and J. F. Piatt. 1990. Body size and foraging behaviour in birds. Acta XX Congr. Int Orn.:811-816.

Greenwd, J. G. 1979. Post-rnortem shrinkage of Dunlin Calidris alpina skins. Bull. of the British Omithol. Club 99:143-145.

Groves, S. 1978. Age-related differences in Ruddy Turnstone foraging and aggressive behaviow. Auk 95:95- 103.

Harestad, A. S. and F. L. Bunnell. 1979. Home range and body weight-a reevaluation. Ecology 6û:389-402.

Harrington, B.A. 1982. Morphometic variation and habitat use of Semipalmated Sandpipers during a migratory stopover. J. Field Omithol. 53:258-262

Heitmeyer, M. E., D. P. Connely, and R. L. Pederson. 1989. The Central, Imperia1 and Coachella valleys of California, p. 475-505. In: L. M. Smith, R. L. Pederson and R. M. KUninski [eds.], Habitat management for migrating and winter waterfowl in North Amenca. Texas Tech. Universtiy Press, Texas.

Heppleston, P.B. 197 1. The feeding ecology of Oystercatchers Haematopus osiralegus L. in winter in northem Scotland. I. Anim. Ecol. 41 :65 1-672.

Hestbeck, J. B., J. D. Nichols, and R. A. Malecki. 1991. Estimates of movement and site fidelity using mark-resight data of wintering Canada Geese. Ecology 72:523-533.

Hicklin, P. W. and f .C. Smith. 1979. The diets of five species of migrant shorebirds in the Bay of Fundy. Proç. N.S. Inst. Sci. 29:483-488.

Hill, D., S.P. Rushton, N. Clark, P. Green, and R. Prys-Jones. 1993. Shorebird communities on British estuaries: factors affecthg cornmunity composition. J. Appl. Ecol. 30:220-234.

Hixon, M. A., F. L. Carpenter, D. C. Paton. 1983. Territory area, flower density, and time budgeting in hummingbirds: an experimental and theoretical analysis. Am. Nat 122:366- 391.

Hockey, P. A. R, Tutpie, J. K. & Velbquez, C. R. 1998. What selective pressures have driven the evolution of deferred northward migration by juvenile waders? J. Avian Biol. 293325 -330.

Holmes, R. T. 1970. Differences in population density, territoriality, and food supply of Duadin on arctic and subarctic tundra, p. 303-3 19. En A. Watson, Animal populations in dat ion to their food tesources. Oxford Univ. Press, Oxford.

Holmes, R. T. 1971. Latitudinal differences in the breeding and molt schedules of Alaskan Red-backed Sandpipers (Calidris alpina). Condor 73 :93-99.

Hooge, P. N. and B. Eichenlaub, 1997. Animal movement extension to Arcview, vers. 1.1. Alaska Biol. Sci. Ctr., U. S. Geol. Survey, Anchorage, AK.

Hôtker, H. 2000. When do Dunlins spend high tide in flight? Waterbirds 23:482-485.

Johnson, D. H. 1980. The cornparison of usage and availability measurements for evaluating tesaurce preference. Ecology 61 55-71.

Kaiser, G.W. and M. Gillingham. 1985. Some notes on seasonal fluctuations in the weight of Duniin Calidris aipina on the Fraser River Delta, British Columbia. Wader Study Group Bull. 3 1 :46-48.

Kaiser, T. J. and T. L. Bucher. 1985. The consequences of reverse sexuia size dimorphism for oxygen consurnption, ventilation, and water loss in relation to arnbient temperature in the Prairie Falcon, Fuko mexicanus. Physiol. 2001. 58:748-758.

Ketterson, E. D., and V. Nolan Ji. 1976. Geographic variation and its climatic correlates in the sex ratio of eastem-wintering Dark-eyed Jwos (Junco hyenralis hyemalis). Ecology 57:679693.

Ketterson, E. D., and V. Nolan Jr. 1979. Seasonal, annual, and geographic variation in sex ratio of wintering populations of Dark-eyed Juncos (Junco hyemalis). Auk 96532- 536.

Ketterson, E. D., and V. Nolan Jr. 1983. The evolution of differential bird migration. Current Ornithol. 1 :357402.

Kelly, P. R and H. L. Cogswell. 1979. Movements and habitat use by wintering populations of Willets and Marbled Godwits. Stud. Avian Biol. 2:69-82.

King, J. R. 1974. Seasonal allocation of time and resources in birds, p. 4-85. In A. Paynter, Avian energetics. Nuttall Ornithological Club 15, Cambridge, MA.

Kodric-Brown, A. and J. H. Brown. 1978. lanuence of economics, interspecific cornpetition, and s e x d dimorphism on temtoriality of migrant Rufous Humuiingbirds. Ecology 59:285-2%.

Kus, B. E. 1985. Aspects of the flocking behavior and predator avoidance in wintering shorebirds. Unpubl. PhD. Thesis, UC Davis.

Levings, C. D., and R. M. Thom. 1994. Habitat changes in Georgia Basin: implications for resource management and restoration, pp.330-35 1. In: R. C. H. Wilson, R J. Bearnish, F. Aitkens, and J. Bell [eds.], Review of the marine environment and biota of Strait of Georgia, Puget Sound and Juan de Fuca Strait. Canadian Technicai Report of Fisheries and Aquatic Sciences #1948.

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

Lima, S.L. 1986. Predation risk and unpredictable feeding conditions: determinants of body mass in birds. Ecology 67:377-385.

Lima, S. L. and L. M. Dill. 1989. Behavioural decisions made under the risk of predation: a review and prospectus. Can. J. Zool. 68:619-640.

Littell, R. C., Freund, R. J., and Spector, P. C. 1991. SAS System for Linear Models (3rd Edition). ASA Institute, Cary, NC.

Lougheed, S. C., T. W. Arnold, and R. C. Bailey. 1991. Measurement error of extemal and skeletai variables in birds and its effect on principal components. Auk 108:432-436.

Lowom, J. R. and K. R. Baldwin. 1996. Intertidal and fmland habitats of ducks in the Puget Sound region: a landscape perspecitve. Biol. Conserv. 77:97-114.

Maclean, S. F., Jr., and R. T. Holmes. 1971. Bill lengths, wintering areas, and taxonomy of North American Dunlins, Calidris alpina. Auk 88:893-901.

Marchetti, K. and T. Price. 1989. Differences in the foraging ofjuvenile and adult birds: the importance of developmental constraints. Biol. Rev. 64:s 1-70.

McCloskey, J. T., J. E. Thompson. 2000. Sex-related differences in migration chronology and winter habitat use of Common Snipe. Wilson Bull. 1 12:143-148.

McEwan, E.W., and A. Fan. 1986. Abundance and distribution of benthic invertebrates in the intertidal mudflats of Boundary Bay. Can. Wildl. Serv. Unpubl. Rep.

McEwan, E. W., and P.M. Whitehead. 1984. Seasonal changes in body weight and composition of Dunlin (Calidris alpina). Cm. J. 2001.62: 154-1 56.

MacLean, S. F. and T. R. Seastedt. 1979. Avian territoriality: sufficient resources or interference cornpetition. Am. Nat. 1 14308-3 12.

MacLean, S.F., and R.T. Holmes. 1971. Bill lengths, wintering areas, and taxonomy of North American Dunlins, Calidris alpina. Auk 88:893-901.

Mares, M. A. and T. E. Lacher, Jr. 1987. Social spacing in small mammals: patterns of individual variation. Amer. Zoo!. 27:293-306.

Marra, P. 1999. The role of behavioural dominance in structuring patterns of habitat occupancy in a migrant bird during the nonbreeding season. Behav. Ecol. 1 1 :299-308.

Mawhinney-Gilliland, K.M. 1992. Foraging ecology and body condition of migrant Semipalmated Sandpipers, Calidris pusilla, on three mudflats in the Bay of Fundy Hernispheric Shorebud Reserve, Nova Scotia. MSc. thesis, Acadia University, Wolfville, N.S. 74pp.

McCloskey, J. T., J. E. Thompson. 2000. Sex-related differences in migration chronology and winter habitat use of Common Snipe. Wilson Bull. 1 12: 143.148.

McCullough, D. 1996. Arcview triangulation program developed with the Avenue Programming Language. Delta, B. C.

McEwan, E. W., and A. Fm. 1986. Abundance and distribution of benthic invertebrates in the intertidal mudflats of Boundary Bay. Can. Wildi. Sem. Unpubl. Rep.

McLaren, P. and P. Ren. 1995. Sediment transport and its environmental implications in the lower Fraser River and Fraser Delta. Environ. Canada Publ. DOE FRAP1995-03.

McNab, B. K. 1963. Bioenergetics and the determination of home range size. Am. Nat. 97:133-140.

McNeil, R., P. Drapeau, and J.D. Goss-Custard. 1992. The occurrence and adaptive significance of nocturnal habits in waterfowl. Biol. Rev. 67:38 1-4 19.

Metcaife, N. B. 1984. The effects of habitat on the vigilance of shorebirds: is visibility important? Anim. Behav. 32:98 1-985.

Metcalfe, N. B. and R. W. Fumess. 1984. Changing priorities: the effect of pre-migratory fattening on the tradesff between foraging and vigilance. Behav. Ecol. Sociobiol. 15: 203-206.

Metcalfe, N.B. and S.E. Ure. 1995. Diumal variation in flight performance and hence potential predation risk in smail birds. Proc. R. Soc. Lond. B 261: 395-400.

Minton 1975. Waders of the Wash - ringing and biomeûic studies. Report to NERC of Scientific Shidy G of the Wash Wader Sturage Scheme Feasibility Study. 40 pp.

Momw, A. and R. McNeil. 1991. Time-Activity Budget Of Wilson's and Semipalmateci Plovers In A Tropical Environment. Wilson Bull. 103598-620.

Morris, D. W. 1987. Ecologicai scale and habitat use. Ecology 68:362-369.

Momson, R. 1. G., Y. Aubry, R. W. Butler, G. W. Beyersbergen, G. Donaldson, P. W. Hicklin, V. H. Johnston, R. K. Ross, and C. L. Gratto-Trevor. (in press). Declines in North American shorebird populations. Absûact fiom the Waterbirds meeting, Nov. 2000, Wader Study Group Bull.

Momson, R. 1. G., R. E. Gill Jr., B. A. Harrington, S. Skagen, G. W. Page, C. L. Gratto- Trevor, and S. M. Haig. 2001. Estimates of shorebird populations in North America. Can. Wildl. Serv. Occas. Paper No. 104. Ottawa, Canada. 62 pp.

Mouritsen, K.N. 1992. Predator avoidance in night-feeding dunlins Calidris alpina. Omis Stand. 23:195-198

Mouritsen, K.N. 1993. Diunral and nocturnai prey detection by Dunlins Calidris alpina. Bird Study 40:2 1 2-2 15.

Mouritsen, K.N. 1994. Day and night feeding in Dunlins Calidris alpina: choice of habitat, foraging technique and prey. J. Avian Biol. 2555-62.

Myers, J. P. 198 1. A test of three hypotheses for latitudinai segregation of the sexes in wintering birds. Cm. J. Zool. 59: 1527-1 534.

Myers. 1984. Spacing behaviour of nonbreeding shorebirds, p.27 1-32 1. In: Burger, J. and B. L. Olla [eds.]. Shorebirds: Migration and Fotaging Behaviour. Plenum Press, New York.

Myers, J.P., P.G. Comors and F.A. Pitelka. 1979. Temtory size in wintering Sanderlings: the effects of prey abundance and inûuder density. Auk 9655 1-56 1.

Myers, J. P., S. L. Williams, and F. A. Pitelka. 1980. An experimental analysis of prey availability for Sanderlings (Aves: Scolopacidae) feeding on sandy beach crustaceans. Cm. J. Z00l. 58: 1564- 1574.

Myers, J. P., P. G. Comors, and F. A. Pitelka. 1981. Optimal temtory size and the Sanderlings: compromises in a variable environment. Foraging Behavior: Ecologicai, Ethological, and Psychologicai Approaches. A. C. Kamil and T. D. Sargent. N.Y., Garland STPM Press: 135- 158.

Myers, J. P. and B. J. McCaffery. 1984. Paracas revisited: do shorebirds compete on their wintering ground? Auk 101 : 197-1 99.

Myers, J.P., RLG. Morrison, P.T.Z. Antas, P. Canaveri, B.A. Hanington, T.E. Lovejoy, V. Pulido, M. Sallaberry, and S.E. Semer. 1987. The Western Hemisphere Shorebird Reserve Network. Wader Shidy Group Bull. 49: 122- 124.

Myers, J. P., C. T. Schick, and G. Castro. 1988. Structure in Sanderling (Calidris aiba) populations: the magnitude of intra- and interyear dispersal duting the nonbreeâiig season. P m . Int. Orn. Congr. XiX:604-615.

National Wetlands Working Group. 1988. Wetlands of Canada. Environ. Canada and Polysci. Publ, Inc., Montreal, Quebec.

Orians, G. H. and J. F. Wittenberger. 1991. Spatial and temporai scales in habitat selection. Am. Nat. I37:S29-S49.

Osuji, P. 0.1974. The physiology of eating and the energy expenditure of the nuninant at pasture. J. Range Manage. 27:437-443.

Page, G. 1974. Age, sex, molt and migration of Dunlin at Bolinas Lagoon. Western Birds 5:1-12.

Page, G. and D.F. Hihitacre. 1975. Raptor predation on winteing shorebirds. Condor 77:73-83.

Page, G.W., L.E. Stenzel and C.M. Wolfe. 1979. Aspects of occurrence of shorebuds on a central California estuary, pp. 15-32. In: F.A. Pitelka [ed.]. Shorebirds in marine environments. Stud. Avian Biol. No. 2. Cooper Osnithol. Soc., Allen Press.

Page, G., and R.E. Gill. 1994. Shorebirds in western North America: late 1800s to late 1900s. Stud. Avian Biol. 15: 147-160.

Page, G. W., L. E. Stenzel, and 1. E. Kjelmyr. 1999. Overview of shorebird abunâance and distribution in wetlands of the Pacific Coast of the contiguou United States. Condor 101 :461-471.

Paulson, D. 1993. Shorebirds of the Pacific Northwest. Univ. Wash. Press, Seattle.

Pedlar, J. H., L. Fahrig, and H. G. Memam. 1997. Raccoon habitat use at 2 spatial scales. J. Wildl. Manage. 61: 102-1 12.

Peery, M. 2.2000. Factors affécting interspecies variation in home range size of raptors. Auk 117511-517.

Pienkowski, M.W. 198 1. How foraghg plovers cope with environmental effécts of invertebrate behaviour and availability, pp. 179- 192. In: N.V. Jones and W.F. Wolff [eds.]. Feeding and SuMval Sûategies of EstuarirPe Organisms. Plenum Press, London.

Pienkowski, M.W. 1982. Diet and energy intake of Grey and Ringed plovers, Pluvialis squuturola and Charahius hiuricula, in the non-breeding season. J. Zool., Lond. 197:s 1 1-549.

Pienkowski, M. W., and W. J. A. Dick. 1975. The migration and wintering of Dunlin Calidris alpinu in north-west Afnca Omis Scand. 6: 15 1-1 67.

Pienkowski, M. W. and H. Clarke. 1979. Preliminary results of winter dye-marking on the Firth of Forth, Scotland. Wader Study Group Bull. 27: 16-1 8.

Pienkowski, M.W. C.S. Lloyd and C.D.T. Minton. 1979. Seasonal and migrational weight changes in dunlins. Bird Study 26: 134- 148.

Piersma, T., R. Hoekstra, A. Dekinga, A. Koolhaas, P. Wolf, P. Battley, and P. Wiersma. 1993. Scale and intensity of intertidd habitat use by knots Calidris canutus in the western Wadden Sea in relation to food, friends and foes. Neth. J. Sea Res. 3 1 :33 1-357.

Piersma,T., Van Gils, J., DeGoeij, P., Van Der Meer, J. 1995. Holling's finctional response mode1 as a tool to link the food-finding mechanism of a probing shorebird with its spatiai distribution. J. Anim. Ecol. 64493-504.

Piper, W. H. and R. H. Wiley. 1990. Correlates of range size in wintering white-throated sparrows, Zonotrichia albicollis. Anim. Behav. 40545-552.

Plissner, J. H., L. W. Oring, and S. M. Haig. 2000. Space use of Killdeer at a Great Basin breeding area. J. Wildl. Manage. 64:421-429.

Prater, A. J., J. H. Marchant, and J. Vuorinen. 1977. Guide to the identification and ageing of Holarctic waders. British Trust for Ornithol., Field Guide # 17, Tring, United Kingdom.

Puttick, G. M. 1979. Foraging behaviour and activity budgets of Curlew Sandpipers. Ardea 67: 1 1 1 - 122.

Puttick, G. M. 1981. Sex-related differences in foraging behaviour of Curlew Sandpipers. Omis Scand. 12: 1 3- 17.

Puttick, G. M. 1984. Foraging and activity patterns in wintering shorebirds, p. 203-232. In J. Burger and B. L. Olla [eh.], Shorebirds-migration and foraging behaviour. Plenum Press, New York.

Quammen, M.L. 1982. Influence of subde substrate differences on feeding by shorebirds on intertidal mudflats. Mar. Biol. 71:339-343.

Rands, M. R. W. and J. P. Barkham. 198 1. Factors controllhg within-floçk feeding densities in three species of wading bird. Mis Scand. 12:28-36.

Redpath, S. 1988. Vigilance kvels in preening Dunlin Calidris aipina. Ibis l3O:SS5-557.

Relyea, R. A., R. K. Lawrence, and S. Desmarais. 2000. Home range of Desert Mule D m : testing the body-size and habitat-productivity hypotheses. J. Wildl. Manage. 64:146-153.

Rising, J. D, and K. M. Somers. 1989. The rneasurement. of overall body size in b i s . Auk 106: 666-674.

Robert, M. and R. McNeil. 1989. Comparative day and night feeding strategies of shorebird species in a tropical environment. Ibis 13 1 : 69-79.

Robert, M., R. McNeil, and A. Leduc. 1989. Conditions and significance of night feeding in shorebirds and other water birds in a tropical lagoon. Auk 106:94-101.

Rompre, G., McNeil, R. 1994. Seasonal changes in &y and night foraging of Willets in Northeastem Venezuela. Condor 96:734-738.

Rottenborn, S. C. 19%. The use of coastal agncultural fields in Virginia as foraging habitat by shorebirds. Wilson Bull. 108:783-7%.

Ruiz, G. M., P. G. Cornors, S. E. Grifin, and F. A. Pitelka 1989. Structure of a wintering Dunlin population. Condor 9 1 562-570.

Saab, V. 1999. Importance of spatial scale to habitat use by breediig birds in riparian forests: a hierarchical analysis. Ecol. Appl. 9: 135- 15 1.

Samuel, M. D., D. J. Pierce, and E. O. Garton. 1985. IdentiSling areas of concentrated use within the home range. J. Anim. Ecol. 54:711-719.

Schnute, J., and D. Fournier. 1980. A new approach to length-Erequency analysis: growth structure. Cm. I. of Fish. and Aquat. Sci. 37: 1 337-1 35 1.

Schoener, T. W. 1968. Sizes of feeding temtocies among b i s . Ecology 49: 123-141,

Schoener, T. W. 1 983. Simple models of optimal feeding-territory size: a teconciliation. Am. Nat. 12 1 Ai08429

Scott, I., P. 1. Mitchell, and P. R Evans. 1996. How does variation in body composition affect the basal metabolic rate of birds? Funct. Ecol. 10:307-3 13.

Seaman, D. E., J. J. Millspaugh, B. J. Kernohan, G. C. Brundige, K. J. Raedeke, R. A. Gitzen. 1999. Effects of sample size on kemel home range estimates. J. Wildi. Manage. 63:739-747.

Selman, J. and J. D. Goss-Custard. 1988. Interference between foraging Redshank Tringa totanus. Anim. Behav. 36: 1542- 1554.

Sewell, M.A. 1996. Detection of the impact of predation by migratory shorebirds: an experimental test in the Fraser River estuary, British Columbia (Canada). Mar. Ecol. Progr. Ser. 14423-40.

Sewell, M.A. and R. W. Elner. 2001. Spatial variability in the macrofauna of intertidal flats on Boundary Bay, British Columbia. Can. Wildl. Sem. Tech. Rep. No. 365, Ottawa.

Shepherd, P. 2001. Status and conservation of Dunlin (Calidris alpina) in Canada. Bird Trends #8, Can. Wildl. Serv. Publ., Ottawa, Canada.

Shuford, W. D., G. W. Page, and L. E. Stenzel. 1998. Patterns and dynamics of shorebird use of California's CentraI Valley. Condor 100:227-244.

Sitters, H. P. 2000. The role of night-feeding in shorebirds in an estuarine environment with specific reference to mussel-feeding oystercatchers. Ph.D. Thesis, University of Oxford, England.

Skagen, S.K. and F. L. Knopf, 1993. Toward conservation of midcontinental shorebird migrations. Consew. Biol. 7533-54 1.

Smith, P. C. and P. R. Evans. 1973. Studies of shorebirds at Lindisfame, Northumberland. 1. Feeding ecology and behaviour of the Bar-tailed Godwit. Wildfowl 24:135-139.

Smith, B. D., and G. S. Jamieson. 1989. Growth of male and female Dungeness crabs near Tofino, British Columbia. Trans. Am. Fish. Soc. 1 18:55&563.

Smith, B. D., and L. W. Botsford. 1998. Interpretation of growth, mortality, and recruitrnent patterns in size-at-age, growth increment, and size îÎequency data, p. 125- 139. In G. S. Jamieson and A. Campbell [eds.], Proceedings of the North Pacific Symposium on Invertebrate Stock Assessment and Management. Can. Spec. hbl . Fish. and Aquat. Sci. 125.

Smith, K.W., LM. Reed and B.E. Trevis. 1992. Habitat use and site fidelity of Green Sandpipers Tringa ochropus wintering in southern England. Bird Study 39: 155-164.

Spaans, A. L. 1977. Are starlings faithful to their individual winter quarters? Ardea 6483-87.

Speth, J. 1979. Conservation and management of coastal wetlands in California. Stud. Avian Biol. 2:151-155.

Stillman, R. A., J. D. Goss-Custard, and R. W. G. Caldow. 1997. Modelling interference from basic foraging behaviour. J. Anim. Ecol. 66:692-703.

Stillman, R. A., J. D. Goss-Custard, A. D. West, S. E. A. Le V. Dit Durrell, R. W. G. Caldow, S. McGrorty, and R. T. Clarke. 2000. Predicting mortality in novel environments: tests and sensitivity of a behaviour-based model. J. Appl. Ecol. 37564- 588.

Storch, 1. 1995. Annual home ranges and spacing patterns of Capercaillie in Central Europe. J. Wildl. Manage. 59:392-400.

Strauch, J. G. 1967. Spnng migration of Dunlin in interior western Oregon. Condor 69:21&212.

Summers, R.W., M. Nicoll, L.G. Underhill and A. Petersen. 1988. Methods for estimating the proportionsof Icelandic and British Redshanks Tringa totanirs in mixed populations wintering on British coasts. Bird Study 35: 16% 180.

Sutherland, T. F., P. C. F. Shephecd, and R. W. Elner. 2000. Predation on meiofaunal and macrofaunal invertebrates by Western Sandpipers (Calidris mauri): evidence for selection and dual feeding modes. Mar. Biol. 137:983-993.

Sutherland, W. J. and L. M. Gosling. 2000. Advances in the study of behaviour and their role in conservation, pp. 3-9. In: L. J. Gosling and W. J. Sutherland [eds.]. Behaviour and Conservation. Cambridge University Press, Cambridge.

Swinbanks, D.D. 1979. Environmentai factors conirolling floral zonation and the distribution of burrowing and tube-dwelling organisms on Fraser Delta tidal flats, British Columbia. PhD. thesis, Univ. of British Columbia, B.C. 275 pp.

Taitt, M. J. 1997. Wildlife Monitoring of Farmland Cover Crops in 199516. Report prepared for the Delta Farmland and Wildlife Trust, Delta, B. C.

Thibault, M and R. McNeil. 1994. Dayfnight variation in habitat use by Wilson's Plovers in northeastem Venezuela. Wilson Bdl. 106:299-3 10

Townshend, J.D. 198 1. The importance of field feeding to the SUVival of wintering male and female Curlews Numenius arquata on the Tees esniary. In: N.V. Jones and W.J. Wolff [eds.]. Feeding and survival strategies of estuarine organisms. Plenum Press, New York.

Townsend, D.J., P.J. Dugan, and M.W. Pienkowski. 1984. The unsociable plover: use of intertidal areas by Grey Plovers. In: P.R. Evans, J.D. Goss-Custard, and W.G. Hale [eds.]. Coastal Waders and Wildfowl in Winter. Cambridge U. Press, Cambridge.

Tripp, K. J. and J. A. Collazo. 1997. Non-breeding temtoriality of semipalmated sandpipers. Wilson Bull. 109: 630-642.

Turpie, J. K. 1995. Non-breeding temtoriality: causes and consequences of seasonal and individual variation in grey plover Pluvialis squatarola behaviour. J. Anim. Ecol. 64:429- 438.

Twpie, J.K. and P.A.R. Hockey. 1993. Comparative diurnal and nocturnal foraging behaviour and energy intake of premigratory Grey Plovers Pluvialis squatarola and Whimbrels Numenius phaeopus in South Afnca. Ibis 135: 156- 165.

Turpie, J.K. and P.A.R. Hockey. 1996. Foraging ecology and seasonal energy budgets of estuarine Grey Plovers Pluvialis squatarola and Whimbrels Numenius phaeopus at the southem tip of Africa. Ardea 8457-74.

Van der Have, T.M., E. Nieboer, and B.C. Boere. 1984. Age-related distribution of Dunlin in the Dutch Wadden Sea. In: P.R. Evans, J.D. Goss-Custard, and W.G. Hale [eds.]. Coastal Waders and Wildfowl in Winter. Cambridge University Press, Cambridge.

Vandermeer, J. and 1. Perfecto. 1997. The agroecosystem: a need for the conservation biologist's lens. Cons. Biol. 11591-592.

Velasquez, C.R. and P.A.R. Hockey. 199 1. The importance of supratidal foraging habitats for waders at a south temperate estuary. Ardea 80:243-253.

Walpole, R. E., R. H. Myers, and S. Myers. 1998. Probability and statistics for engineers and scientists. 6th edition. Prentice Hall Press, Upper Saddle River, NJ.

Wamock, N.D. 1994. Biotic and abiotic factors affecting the distribution and abundance of a wintering population of Dunlin. Unpubl. PhD. Thesis, UC Davis and San Diego State, Califomia. 145pp.

Warnock, N.D., G.W. Page, and L.E. Stenzel. 1995. Non-migratory movements of Dunlin on their California wintering grounds. Wilson Bull. 107: 13 1-139.

Warnock, N. D., and R. E GiU. 1996. Dunlin (Calidris alpina). In: A. Poole and F. Gill [eds.]. The birds of North America, No. 203. The Academy of Naturai Sciences, Philadelphia, and The American Omithologists' Union, Washington, DC.

Wamock, N., G. W. Page, and L. E. Stenzel. 1995. Non-migratory movements of Dualin on their Califomia wintering grounds. Wilson Bulletin 107: 13 1-1 39.

203

Wamock, N., G. W. Page, and B. K. Sandercock. 1997. Local survival of Dunlin wintering in California. Condor 99:9069 15.

Wamock, S.E and J.Y. Takekawa. 1996. Wintering site fidelity and movement patterns of Western Sandpipers Calidris mauri in the San Fransisco Bay estuary. Ibis 138:160-167.

Weber, L. M. and S. M. Haig. 1996. Shorebird use of South Carolina managed and natural coastal wetlands. J. Wildl. Manage. 60:73-82.

Whitfield, D.P. 1985. Raptor predation on wintering waders in southeast Scotland. Ibis 127544558.

Whittingharn, M. J. 1996. The use of radio telemetry to measure the feeding behaviour of breeding European Golden-Plovers. J. Field Omithol. 67:463-470.

Willson, M. F. 1971. Seed selection in some North American finches. Condor 73:415- 429.

Worton, B. J. 1995. Using Monte Carlo simulation to evaluate kemel-based home range estimators. J. Wildl. Manage. 59:794-800.

Yates, M. G., J. D. Goss-Custard, S. McGrorty, K. H. Lakhani, S. E. A. Le V. Dit Durell, R. T. Clarke, W. E. Rispin, 1. Moy, T. Yates, R. A. Plant, and A. J. Frost. 1993. Sediment characteristics, invertebrate densities and shorebird densities on the inner banks of the Wash. J. Appl. Ecol. 30599-614.

Yates, M. G., J. D. Goss-Custard, and W. E. Rispin. 1996. Towards predicting the effect of loss of intertidai feeding areas on overwintering shorebirds (Charadrii) and shelduck (Tadorna radorna): refinements and tests of a mode1 developed for the Wash, east England. J. Appl. Ecol. 33:944-954.

Ydenberg, R. C. and J. R. Krebs. 1987. The tradeoff between temtorial defense and foraging in the Great Tit (Parus major). Amer. Zool. 27:337-346.

Zwarts, L., A.-M. Blomert, and R. Hupkes. 1990. Increase in feeding time in waders prepariag for spring migration fiom the Banc dYArguin, Mauritania. Ardea 78:237-256.

Zwarts L., A.-M. Blomert, B. J. Ens, R. Hupkes, T. M. van Spanje 1990. Why do waders reach high feeding densities on the intertidai flats of the Banc d'Arguin, Mauritania? Ardea 78:39-52.