Canada. du Canada. Bibliothèque nationale. National Library. Acquisitions et. Bibliographie Services. Acquisitions and. ûttawaûn KtAON4 Canada

Transcription

1

2

3 Natonal Lbrary Acqustons and Bblographe Servces Bblothèque natonale du Canada Acqustons et servces bblographques 395 Wellngton Street Ottawa ON KI A ON4 395, rue Wdlngtm ûttawaûn KtAON4 Canada Canada The author has granted a nonexclusve lcence aowng the Natonal Lbrary of Canada to reproduce, loan, dstrbute or sel1 copes of ths thess n mcroform, paper or electronc formats. The author retans ownershp of the copyrght n ths thess. Nether the thess nor substantal extracts f?om t may be prnted or othemse reproduced wthout the author's permsson. L'auteur a accordé une lcence non exclusve permettant à la Bblothèque natonale du Canada de reprodure, prêter, dstrbuer ou vendre des copes de cette thèse sous la forme de mcrofche/fh, de reproducton sur paper ou sur format électronque. L'auteur conserve la proprété du drot d'auteur qu protège cette thèse. N la thèse n des extrats substantels de celle-c ne dovent être mprmés ou autrement reproduts sans son autorsaton. Canada

4 INVESTIGATION OF CAUSES AND EFFECIS OF PREDATION BY HERRING (Lmus mgentatus) AND GREAT BLACK-BACKED GULLS (L. mmnus) ON BLACK-LEGGED KITTIWAKES (Rsu hdpctyla) ON GULL ISLAND, NEWFOUNDLAND A thess submtted to the School of Graduate Studes n partal fulfllment of the requrements for the degree of Master of Scence Department of Bology Mernoral UNversty of Newfoundland May 2000 St. John's Newfoundland Canada

5 ABSTRACT OF THE THESIS In prevous shdes t has been observed that herrng mgentatus) and great black-backed gulls (L. murnus) depredated breedng black-legged kttwakes (Rssa hdactytyla) Uat nest dong the southeaskm coast of Newfoundland, Canada. However, the causes and effects of large gull predaton on kttwakes was never extensvely nvestgated nor quantfed. In ths study, herrng gu1 and great black-backed gull predaton on black-legged kttwakes at Gull Island, southeastern Newfoundland was quantfed at four study plots n relaton to the tmng of the annual spawnng arrval of capeln (Mallotus vzlosus), the sue of kttwake sub-colones (number of nets), kttwake nest-ste characterstcs, and wnd condtons. 1 also nvestgated the mpact of large gul predaton on kttwake breedng performance durng 1998 and I compared large gulls' predaton attempt frequency among three perods: before mean gull hatchng, between mean gull hatchng and the anval of capeln, and folowng capeln arrval. In both years, the frequency of gull predaton attempts on kttwakes dflered sgnfmntly arnong the three perods, wth hghest levels of predaton occurrng after gull chcks hatched but before capeln arrval. Overall gull predaton attempt levels were lower n 1999, when capeln spawned earler, than n Melane Massaro - Lmgc gull predaton a bacù-kgged kttwakes

6 Nestng densty and the locaton on the clff were kttwake nest-ste characterstcs that affected sgnhcantly the rsk of predaton. Breedng success (number of successfu nests) was nfluenced by nestng densty and ledge wdth- Addtonally, 1 found that both rsk of predaton and breedhg success vaned sgnfcantly among plots. Indvdual kttwake nests at the smallest plot experenced a hgher probablty of attack by large gulls than nests at!arger plots. Hence, the percentage of faled nests was hghest at the smallest plot and decreased as the sze of the plots ncreased. Regatdess of wnd condtons both gull speces attacked nest stes located on upper parts to a hgher lkelhood than stes located on mddle and lower parts of the dffs. However, durng calrn condtons, roofs over nest stes reduced the rsk of predaton by hemng gulls, whereas stes located on narrow ledges were less lkely to be attacked by great black-backed guk Durng wndy condtons, nestng densty affected whch stes were attacked by great black-backed gulls. The level of gull predaton behavour was sgnhcantly correlated wth the percentage of kttwake eggs and chcks that dsappeared wthn a week. 1 estmated that 43% of kttwake eggs and ch* at G d Island were taken by gulls n 1998 and 30% n My resuts demonstratec that kttwakes have been ndrdy (through naeased predaton by gds) affected by the deayed arrva1 and lower abundance of capeln, and that kttwake nest-ste Melme Massmo - Lmge guü pttdatm on black-iegged kttwaks

7 characterstcs, and the sze of a sukolony were sgnucantly correlated wth the rsk of depredaton.

8 FOR MY MUM

9 ACKNOWLEDGEMENTS Frstly, 1 would lke to thank my supervsor John Chardne for provdng moral and academc support throughout ths study, for always tndng the for dscussons when 1 needed help and for fnancal support receved through hs grants. Greg Robertson helped n many ways. Mostly 1 thank hm for sharng hs nvaluable statstcal knowledge, logstc support n the feld, and for hs good advce n many occasons. 1 thank my CO-supervsor, Ian Jones for helphl comments on an earler draft of ths thess and for fnancal support. 1 am deeply ndepted to my feld assstants, Sarah Boyne, Dave Ffeld, Dedrec Grecan, and Lama Penney, who spent many long hours n observaton blnds, endurng cold and many tcks, to record kttwake and gui behavour. Ther enthusasm and good hurnor made both seasons on Gd Island unforgettable. Many thanks also to al volunteers helpng out n the feld. Thanks to Shauna BaUe for provdng nformaton on the puffn det n Thanks to the members of my cornmttee, Anne Storey and Ted Mller, who mproved ths study wth ther constructve crtcsm n earler stages of the study as well as on an earler draft of ths thess. To Keth Ems, Gad Fraser, Stefan Garthe, Grant Glchrst, Scott Gllland and Mark Hpfner 1 am grateful for helpng developng deas and lookng at earer manuscrpts.

10 Wthout the logstc help and constant support of Joe and Loyola O'Bren of Brd Island Charters n Bay Bulls ths study would have not be possble. They provded transportaton o Gd Island, but more mportant than ths they provded lots of hurnor and cornfort. 1 wsh to thank Mernoral Unversty of Newfoundland for supportng my graduate work wth a scholarshp, n partdar thanks to Murray Colbo, Head of the Bology Department, for hs support. 1 also uwk the Canadan Wldlfe Servce and the Adantc Cooperatve Wldfe Ecology Research Network for fnancal and logstc support, the Department of Toursm, Culture and Recreaton of Newfoundland and Labrador for gvng me a permsson to work on G d Island. 1 would lke to thank the people who 1 saw every day at the Unversty for helpng trelessly wth cornputer, photographc and admnstratve problems: Peter Earle, Chrstne Everson, Roy Fcken, Ga Kenny, Shrley Kenny, Wendy McEvoy and Shena Qunoh Lastly, 1 thank my parents Edth and Horst Neumann for ther constant encouragement and mord support even when I reached for the stars, as wd as Gerd and Hed von Wahert for helpng to fnd and dscover the wonderfu fascnatng world of Bology. Thanks everyone. Melune Massm - Large gull preûatùm on black-leggcd kttwpkts v

11 .- TABLE OF CONTENTS f-gsha%f the thess I &-.r-rwrm.rr-- f Dedcaton Acknowledgements Table of Contents t t ---"J v 1 ~st of Tables I 1 Lst of Fgures xv 1 I Lst of Appendces x f xv? I M I RI : f j Introducton References [ Co-authorshp Statement 10 1! CHAMER2 Delayed capeln (Mallotus vllosrrs) avaablty nfluences large gull 1 predatory behavour on black-legged kttwakes (Rasa tdactyln), 1 I causng a reducton n kttwake bmdng success 2.1. Abstract 2.2. Introducton 13 1 t Melane Massrno - Lmge gd pfe&tùm on biack-legged kttwakrs v

12 2.3. Methods! I Study locaton 16 1 r.-" Predaton behavour frequency f t Breedng success b-! Tmng of gull hatchng 19 I -----! ".-"-.--- ~! f Capeln amval Iuu Statstcal analyses ,,.... w... ~ ~ m Resuts j Tmng f :,-.",,...,,",~ I Gull predaton attempt rates and dsappearance of f kttwake eggs and chcks 32 1 ~"..H...uw~"" The effect of gull predaton on kttwake breedng f SUCC~SS 35 L - I 2.5. Dscusson 37!"...""..-,, f nter- year clfferences n gul predaton rates! "."""""~ Gd predaton rate dferences among study plots The mpact of gull predaton on ktüwake b performance 42 / :- Melane Massrno - Large gu pndnton on bzack-icgged ktfrpakes x

13 p.! 2.6. References " t Relatonshpe between bladc-legged kttwake nest-ste characterstcs and susceptblfy to predaton by large gulls 53 1 I & - $ Introducton 3! 3.3. Methods L-I-.-I-."-"-."-.. t".,.,-.! Study p10s I! / Kttwake nest predaton 59 ~ m m c.. " l «~ ~ ~.. w w w w. w. ~ Nest-ste charac ters tcs w Wnd condtons 65 1 f Breedng success of kttwakes Statstcal analyses -----TA 1! I 3.4. Results Dffaences n guil predaton rsk and kttwake breedug success among plots ncludng all nest stes Gu11 attacks and nest-ste characterstcs n p ~..."-..H.UU...H Kttwake breedng success and nestîte char acte Cornparson of foragng decsons of henng gulls and great black-backed gulls n relaton to wnd condtons f

14 ,...".-.II 3.5. Dscusson. +...~..-~u...cu..~ The effect of plot sze and n n x ; predaton and kttwake breedng success Relatonshps between kttwake nest-ste I characterstcs, gull predaton and kttwake breedng success :"".-."...-ruuuiic.wi l Relatonshps of wnd condtons and foragng decsons of gulls 81 I 3.6. References 85 1 L.. u m ". u m. m w r +. r r r s ~. - ~ ~ ~. u. c.. ~ ~ u u ~ ~ u ~ y u w w ~. f CHAPTER4 1 1 Fnal Dscusson and Condusons 91 1 f.. m......"hiuuu1i.""".œ- I 4.1. References 77 t I 80 j 95j I

15 '..UUl"".".""."..."."""." I Table 21 Generalzed lnear mode1 of factors nfluencng rates of herrng and great black-backed gull predaton attempts on kttwakes on Gul Island n 1998 and I -..""-"- -- Table 2.2 Breedng success of black-legged kttwakes on Gd Island I C.-p"- Table 3.1! j-.- I f I ; ; n 1998 and Defnton of the upper, mddle and lower part for four study plots (SI, PZ, N4 and S5) on Gull Island n 1998 and The dstance of a kttwake nest to the upper edge of a clff defned where a nest was Iocaed relatve to the upper edge of a dff 63 Table 3.2 The number of actve kttwake nests (2 1 egg was lad) : y.- j Table 3.3 fallng wthn each level of nest-ste characterstc, for each of the four study plots on G d Island n 1998 and Nest numbers, percent nests attacked by herrng and great black-badced gulls and percent nests successful n rasng at Ieast one ch& n four kttwake plots on Gd Island n f 1998 and Melane Massaro - Large gull predatan on bkk-kgged ktfbkes x

16 ,"...UH"U--...-IIULj Table 3.4 :. characterstcs that reduced the rsk of predaton attacks by herrng or great black-backed guls on G d Island n 1998 and 1999 (n = 300 randomly chosen stes) 72 m m Table Generalzed lnear mode1 of kttwake nest-ste I 1! J-."CI.U.UI.IUII... Table 3.6 Generalzed lnear model of kttwake nest-ste characterstcs that duenced the breedng success of kttwakes on Gull Island n 1998 and 1999 (n = f randomly chosen stes) 73 Four generalzed lnear models, testng whether certan kttwake net-ste characterstcs nfluenced whch stes were attacked by (1) hehg guk (= ERG) durng calm condtons, (2) great bladc-backed gulls (= GBBG) durng calm condtons, (3) herrng gulls durng wndy condtons, and (4) great black-backed guîls durng wndy condtons 76 1 t f I I! 1 f 4 I f I

17 LIST OF FIGURES r ~ Tmng ofbreedng ~ of grat ~ black-backed ~ gulls (GBBG), ~ ~ -. f herrng gulls (HERG) and black-legged kttwakes (BLKI) on Gull Island and tmng of capeln arrva1 n Wtless Bay n 1998 and Dotted lnes ndcak the perod when brds ncubated eggs and sold lnes d dbe when brds had chcks. Mean hatchng dates for al three brd speces and mean frst-egg layng dates for kttwakes are ndcated by bg crcles. The three nta-seasonal perods for each year are reported wth ther duraton the (d = days) 25 f Fgure 2.2 Mean ) gw - d - O; black-legged kttwakes at four shdy plots on Gull Island...P-" t wthn three ntra-seasonal perods of 1998 and The frst ntra-seasona perod (1) lasted from the begnnng of the season untl mean herrng gull hatchng, the second ; perod (2) from mean herrng gu hatchng untl capeln amval, and the thrd perod (3) from capeln arrva1 untl the end of the season r Fgure 2.3-ton attempmempts/ h o u r ~ o ~ f black-backed gus (GBBG) and herrhg gulls (HERG) on black-legged kttwakes on Gd Island wthn three ntra- seasonal perods of 1998 and The frst ntra-seasonal Melane Massaro - b ge guli pndptm un blrrck-leggcd ùttutpkes xv

18 Tm- t,. -.. perod (1) lasted from the begnnng of the season untd! mean herrng gull hatchng, the second perod (2) from mean herrng gul hatchng untl capeh arrval, and the thrd perod (3) from capeln arrva1 untl the end of the t."l...uiii...p Fgure 2.4 ( ~ ) t a b lkttwake e eggs and chcks 2 (number of kttwake eggs or chcks/total nurnber of eggs lad), (8) the percentage of kttwake eggs and chdo that dsappeared and (C) mean gull predaton attempt rates on Gull Island n each week of the year n 1998 (sold he) and 1999 (slashed lne). The vertcal lnes represent capeln : arrva on 5 Juy 199û (sold vertcal lne) and 26 June 1999 t (slashed vertcal lne) 34 L u u u... w r r ~.. ".. I.I -..-

19 , ".- ~ ~ ~ e n d x ~ ~ ~ ~ t l e s s Bay, ~ e w f o u n d l ~, ' s h o wt n ~ locaton of the fout kttwake study plots and the weather staton n 1998 and ! Appendx 2 Kttwake nest-ste characterstcs for each nest ste at four t study plots (SI, P2, N4 and 55) on Gu Island n 1998 and I ;.".-- 2 AppendDt 3 : Kttwake nest-ste locatons for al stes at four study plots (SI, P2, N4 and S5) on GuU Island n 1998 and Mzlmzè Mnssaro - Large gull pndatun on back-kgged kttroakcs xv

20 CHAPTER 1 Introducton In marne ecosystems seabrds are toppredators that feed on marne fsh, squd, and nvertebrates and t has been shown that changes n abundance of marne prey speces nfluence top-predators and ther breedng performance. (e.g. Barrett and Furness 1990; Hamer et al. 1991; Monaghan et al. 1994; Barrett and Krasnow 19%; Harrs and Wanless 1997; Boersrna 1998; Bryant et al. 1999). Herrng gulls (La rus argenta tus), great blac k-backed gulls (L. mannus) and black- legged kttwakes (Rssa trdnctyln) are surface-feedng brds that are unable to pursue ther prey under water and depend on prey to corne up to the water surface. Surface-feedng seabùds appear to be more susceptble to changes n marne food webs than pursut dvers, sud as puffns and murres (Bard 1990; Barrett and Krasnow 19%; Regehr and Rodway 1999). In Newfoundland, Canada, such a change n abundance of a key marne prey speces for seabrds occurred durng the 19%. Capeln (Mallotus allosus) mgrate nshore from offshore feedng grounds to spawn on kaches along the coast of Newfoundland each sp~g (Templeman 1948). Spawnng capeln are an mportant food resource for seabrds breedng along the Atlantc coast of Newfoundland and Labrador (Burger and Patt 1990; Brown and Nettleshp Melane Mnsssmo - LPge pl1 pdutm on badc-kgged kttzvakes 1

21 19M; Perott and hett 1987; Bryant and Jones 1999). Due to below-normal sea temperatures, the tmng of peak capeln beach spawnng bas ken delayed by approxmately four weeks snce 1991 (Shackel et al. 1994; Themault et al. 1%). The lower water temperatures also affectec maturaton causng reductons n sze of spawnng capeln (Carscadden et al. 1997). The shft n tmng of nshore spawnng of capeh can have a devastatng effect on the breedng performance of seabrds (Regehr and Rodway 1999; Hpfner et al. 2000). As a result of ther flexble foragng behavour, large gulls, such as herrng and great black-backed gulls, benefted frorn dscarded fsh waste by ndustral fsheres (Fumess et al. 1992; Garthe et al. 19%). Ths extra-abundant food resource easly avalable to large guus caused gull populatons to ncrease markedly durng ths century (e.g. Kadlec and Dnvy 1%8; Furness and Monaghan 1987). The coliapse of the northem cod (Gadus morhua) stocks n Newfoundland waters resdted n a moratorum that has essentaly stopped the commercal cod fshery snce The Eastern Canadan Groundfsh Moratorum has lkely decreased the opportuntes for gulls to feed on fsh offal (Regehr and Montevecch 1997). The large-scale reducton of the gound- fsheres (Hutchngs and Myers 1994) and the shft n the tmng of capeh spawnng have forced gulis to seaxch for alternatve food resources. In Newfoundland, large gulls preyed upon adult Atlantc puffns (Fraterculu - Melune Masssmo - Lnrge gull prrdatm on back-legged kttwaks 2

22 arctca), adult leach's stom-petrels (Oceanodmma leumh) and eggs of black- legged kttwakes (Nettleshp 1972; Russell and Montevecch 19%; Regeh and Montevecch 1997; Stenhouse and Montevecch 1999). In contrast to puffns and storn-petrels, black-legged kttwakes nest on vertcal clffs rather than n burrows. Hence, the offsp~g kttwakes are vsudy 'avalable' to large gulls as prey. Qff-nestng n brds evolved as an adaptaton aganst predators, manly mammalan speces (Cullen 1957; Brkhead et al. 19s). Recent studes suggest that clff-nestng also protected breedng thck-bled mues (Ura lomva) aganst predaton by glaucous gulls (L. hypmbo~eus) to a certan degree (Glchrst and Gaston 1997; Glchrst et al. 1998). Nest-ste characterstcs and breedng densty nfluenced under whch wnd condtons glaucous gulls were able to forage successfuly on murre eggs (Glchrst et al. 1998). Compared to murres, black-legged kttwakes buld dstnct nests wth vegetaton and hence the breedng densty s lower. Kttwakes also breed on narrower ledges than mures (Squbb and Hunt 1983). The sze and densty of kttwake sukoiones, as well as fne-scale nest-ste characterstcs may reduce the rsk of predaton by large gulls on kttwakes. Large gulls may change ther foragng tactcs accordng to wnd condtons (Glchrst and Gaston 1997). The objectves of my study were to test whetk delayed capeln avalablty Muences large gull predatory behavour on black-legged

23 kttwakes and whether t affects the breedng performance of kttwakes (Chapter 2). In the thhd chapter 1 examne whether there are any relatomhps between kttwake nest-ste characterstcs and susceptblty to predaton by large gulls. A fnal dscusson of the results of chapters 2 and 3 and general conclusons are presented n Chapter References Bard PH (1990) Influence of abotc factors and prey dstrbuton on det and reproductve nccess of three seabxd speces n Alaska. OMs Scand 21: Barrett RT, Furness RW (1990) The prey and dvng depths of seabrds on Homoy, North Norway &r a decrease n the Barents Sea capeh stocks. Oms Scand 21: Barrett RT, Krasnov W (19%) Recent responses to changes n stocks of prey speces by seabrds breedng n the southem Barents Sea. ICES J Mar Sc 53: 7l3-722

24 Brkhead TR, Greene E, Bggns JD, Nettleshp DN (1985) Breedng ste characterstcs and breedng success n Thck-blled Mmes. Can J : Boersma PD (1998) Populaton trends of the Galapagos Pengun: Impact of El Nno and La Nf. Condor 100: Brown RGB, Nettleshp DN (1984) Capeln and seabrds n the northwest Atlantc. In: Nettleshp DN, Sanger GA, Sprnger PF (eds) Marne brds: thex feedng ecology and commercal fsheres relatonshps. Specal Publcaton, Canadan Wldlfe Servce, Ottawa, Canada, p Bryant R, Jones IL, Hpfner JM (1999) Responses to danges n prey avalablty by Common Mme and Thck-blled Mmes at the Gamet Islands, Labrador. Can J Zoo1 77: Bryant R, Jones IL (1999) Food resource use and det overlap of Common and Thck-blled Mmes at the Gannet Islands, Labrador. Waterbrds 22: Burger AE, Patt JF (1990) Flexble tme budgets n breedng common murres: buffers aganst varable prey abundance. Stud Avan B01 14: 7l-M Carxadden F, Nakashma BS, Frank KT (1997) Effects of fsh length and temperahue on the tmng of peak spawnng n capeln (MnZlotus vllosus). Can J Fsh Aquat Sc 54: Mehe Masssmo - Large gvll predntm un blnck-kgged krrwakes 5

25 Cullen E (1957) Adaptatons n the Kttwake to dff nestng. Ibs 99: Fumess RW, Monaghan P (1987) Seabrd Ecology. New York, Chapman dr Hall Fumess RW, Ensor K0 Hudson AV (1992) The use of hhery wask by gull populatons around the Brtsh Isles. Ardea 80: Carthe S, Camphuysen KCJ, Fumess RW (19%) Amounts of dscar& by commercal fsheres and ther sgnfcance as food for seabuds n the North Sea. Mar Eco1 Pmg Ser 136: 1-11 Glchrst HG, Gaston AJ (1997) Effects of murre nest ste characterstcs and wnd condtons on predaton by glaucous guk. Can J Zool 75: Glchrst HG, Gaston AJ, Smth JNM (1998) Whd and prey nest stes as foragng constrants on an avan predator, the glaucous gull. Ecology 79: Hamer KC, Fumess RW, Caldow RWG (1991) The effects of changes n food avalablty on the breedng ecology of great skuas Gztharacta skua n Shetland. J Zool (Lond) 223: Harrs MP, Wanless S (1997) Breedng success, det, and brood neglect n the kttwake (Rssa hda~tyia) over an Il-year peroà. ICES J Mar Sc 54: Melane Mnsssmo - Large pl1 pdaton a blrtck-hggcd kttmakes 6

26 Hpfner JM, Adams PA, Bryant R (2000) Breedng success of Black-legged Kttwakes, Rssa frdactyla, at a colony n Labrador durng of low capeln, Mallotus allosus, avalablty. Can Feld-Nat 114. In press Hutchngs JA, Myers RA (1994) What can be learned from the couapse of a renewable resource? Atlantc Cod, Gadus rnmhua, of Newfoundland and Labrador. Can J Fsh Aquat Sc 51 : Kadlec JA, Drwy WH (196û) Structure of the New England Herrng Gull populaton. Ecology 49: Monaghan P, Walton P, Wanless S, Uttley JD, Burns MD (1994) Wects of prey abundance on the foragng behavour, dvng effcency, and tme allocaton, of breedng Gullemots Uns aalge. Ibs 136: Netteshp DN (1972) Breedng success of the Common Puffn (Fraterculn ardcn) on dfferent habtats at Great Island, Newfoundland. Ecol. Monographs 42: Perott R, Annett C (1987) Reproductve consequences of detary spechton and swtchng n an ecologca generalst. In: Kaml AC, Krebs JR, PulIarn KR (eds) Foragng behavour. Plenum Ress, New York, p Regehr HM, Montevecch WA (1997) Interactve effects of food shortage and predaton on breedng falure of black-legged kttwakes: ndrect effecf~

27 of fsheres actvtes and mplcatons for nndtor speces. Mar Ecol Prog Ser 155: Regehr HM, Rodway MS (1999) Seabrd breedng performance durng two years of delayed capeln arrva1 n the northwest Atlantc: a multspeces comparson. Waterbrds 22: Russell JO, Montevecch WA (19%) Predaton on adult puffns Fratercula mcta by great black-backed gulls Larus &nus at a Newfoundland colony. Ibs 138: Shackell N, Carscadden JE, Mller DS (1994) Mgraton of pre-spahg capeln (Mallotus vllosus) as related to temperature on the northem Grand Bank, Newfoundland. ICES J Mar Sc 52: Squbb RC, Hunt Jr GL (1983) A comparson of nestng-ledges used by seabrds on St. George Island. Ecology 6L: Stenhouse 1, Montevecch WA (1999) Indrect effects of the avalablty of capeln and fshery dscards: gull predaton on breedng storm-petres. Mar Ecol Prog Ser 1û4: Templeman W (1948) The Me hstory of the capeln (Mallotus vllosus 0. F. Müller) n Newfoundland waters. Bull Newfoundland Gov Lab 17: Melme Masssmo - h ge gull pnàaton a bkk-tegged knwdos 8

28 Themault TW, Schneder DS, Methven DA (1996) The tmng of spawnng n capeln (Mnllotus vliosus Mller) at a coastal locaton n eastern Newfoundland. Polar B01 16:

29 Co-authorshp Statement For both research papers ncluded n ths thess 1 am the prncpal author. 1 have desgned and wrtten the researçh proposal for ths study. 1 spent seveml months n 1998 and 1999 on Gull Island to collect the data presented n ths thess. Fnally, 1 analyzed the data and wrote ths thess. My co-authors, John W. Chardne, Ian L. Jones and Gregory J. Robertson (Chapter 1) were ntellectually nvolved n ths study by gvng advce throughout ths study and makng suggestons to mprove earler drafts of the research papers. Melnne Masssaro - Large pl1 ptcdatun on blpck-kggcd kttzoakes 10

30 Delayed capeln (Mallotus vllosus) avalablty nfluences large gull predatory behavour on black-legged lcttwakes (Rssa trdactyla), causng a reducton n kttwake breedng success To understand causes and effects of varable foragng behavour of large gulls 1 quantfed the mpact of herrng and great black-backed gull predaton on black-legged kttwake breedng success at Gu11 Island, sou theastem Newfoundland n relaton to the tmng of the mual spawnng arrval of capeln durng 1998 and I compared large gulls' predaton attempt frequency among three perods: More mean hehg gull hatchng, between mean gull hatchng and the arrval of capeln, and followng capeln arrval. The frequency of gull predaton attempts on kttwakes dffered spfcantly Ths chper has been accepted for publcaton as a full paper n Canadan Journal of Zoology on 'Delayed capeln (Mdlotus mllosw) avolsbüty nfluences Large gu predatory behavou on black-legged kttwakes (Rssa fnfndactyla), causng a redudon n kttwake breedmg success' (Massaro M, Chardne JW, Jones IL,and Robertson GJ 2000).

31 among the three perods, wth hghest levels of predaton occurrng der gull chcks hatched but before capeln amved. The level of gull predaton behavour was sgdcantly correlated wth the percentage of kttwake eggs and chcks that dsappeared wth a week. 1 estmated that 43% of kttwake eggs and chcks at Gd Island were taken by gulls n 1998 and 30% n Kttwakes have been ndrectly (through ncreased predaton by gulls) affected by the delayed arrva1 and lower abundance of capeln n recent years, underlnng the need to understand mult-speces nteractons when nterpretng the effects of human al teraton of the marne envronment. Melane Masssmo - Large gu1 pndaton on black-kgged kttwakcs 12

32 2.2 Introducton Capeln are small, cmunpolar, schoolng fsh, confned to cool-temperate waters wthn the northem hernsphere (Mc Allster 1%3; Jangaard 1974; Stergou 1989). They spawn on kaches dong the coast of Newfoundlad, Canada, mgratng nshore each sprng from offshore feedng gtounds (Templeman 1948). Spawnng capeh are an essental food resource for many seabrds breedng n Newfoundland. In partcular, durng chck reatng, capeln comprses a large component n the det of common murres (Ura aalge; Burger and Patt IWO), Atlantc puffns (Brown and Nettleshp 1984) and herrng gds (Perott and Annett 1987). Due to below-normal sea temperahres, the tmng of peak capeln beach spawnng has been delayed by approxmately four weeks snce 19% (Shackell et al. 1994; Therrault et al. 19%). Addtonally, Carscadden et al. (1997) showed that cold water temperatures affect maturaton causng reductons n sze of spawnng capeln. In Newfoundland delayed whore spawnng of capeln cm have a devastatng dect on the breedng performance of seabrds, n partdar surface feedng brds, such as black-legged kttwakes, hehg gulls and great bladc-backed gulls (Regehr and Rodway 1999). Large gulls, such as herrng and great black-backed gulls, are detary generdsts, whch feed on marne fsh and nvertebrates as wel as brds and

33 refuse (e.g. Harrs 1%5; 'Ruelfal 1%8; Beaman 1978). As a result of ther flexble foragng behavour, the numbers of large gds have naeased drarnatcdy n Euope and n the northwest Atlantc durng ths century (Kadlec and Dnry 196û; Harrs 1970; Verbeek 1979; Furness and Monaghan 1987; Howes and Montevecch 1993). In partdar, ndustral fsheres offer gulls the oppomuty to feed on fsh offal (e.g. Hudson and Fumess 1989; Furness et al. 1992; Garthe et al. 19%). The Eastern Canadan Groundfsh Moratorum n 1992 has lkely decreased the opportuntes for gulls to feed on fsh offa (Regehr and Montevecch 1997). Large gulls prey on adults and offsprng of several seabrd speces ncudng Atlantc puffns, comrnon and thck-bued murres and black-legged kttwakes (Barrett and Runde 1980; Harrs 1980; Burger and Gochfeld 1984; Schauer and Murphy 1996; Russell and Montevecch 1996; Regehr and Montevecch 1997; Glchrst et al. 1998; Glchrst 1999). Generast foragers are known to swtch det n response to nutrtonal recprements durng the breedng cycle (Perott and Annett 1987) and to changes n prey avalablty (Newton 1993). The largescale reducton of the ground-fsheres n Newfoundland snce 1992 (Hutchngs and Myers 1994; My- and Cardgan 1995) and the shft n the tmng of capeh spawnng have resuted n naeased predaton rates by gulls on obet seabrds (Regehr and Montevecch 1997). After

34 the fshng moratorum great black-backed and herrng guus had an mpact on kttwake breedng success on Great Island n Wtless Bay (Regehr 1994). In 1992, approxmatdy 87% of 416 eggs and n 1993 approwmately 63% of 613 eggs dsappeared, most probably taken by aeral predators (Regehr 1994). In contrast, Maunder and Threlfall(1972) dd not obsenre any kttwake egg predaton by herrng and great black-backed gds on GulI Island, Wtless Bay, and Neuman (1994) observed h-g gulls takng kttwake eggs only twce durng her study just pror to the moratorum n 1990 and Although t s known that herrng and great black-backed gulls prey on black-legged kttwakes, no study has quantfed predaton rates and ther mpact on kttwake breedng success. The man objectve of my study was to document the numbers of gd predaton attempts durng dfferent phases of ther nestng cycle and n relaton to the tmng of capeln arrval. 1 predcted that predaton attempt rates would be hghest when gulls were feedng ther chcks and before capeln had mved. Other specfc objectves were (1) to compare predaton attempt rates arnong several kttwake colones of dfferent sze, (2) to compare the frequency of predaton attempts at dfferent tmes of the day, (3) to compare predaton attempt rates of h&g and great bladc-backed gulls, (4) to compare gull predaton behavour frequency wth the number of Melane Masssmo - LRgc gull pdatm on black-legged kttwrkw 15

35 kttwake eggs and chcks that dsappeared and (5) to mesure the mpact of gull predaton on overall kttwake reproductve success Methods Study locaton The study was conducted on Gull Island (470 16' N, 52' 46' W), part of the Wtless Bay Seabrd Ecologcal Reserve off the southeaskm coast of Newfoundland, Canada. The sland s approxmately 1.6 km long and 0.8 km wde. More than 10,000 pars of black-legged kttwakes breed on clffs dong the edge of Gu11 Island (Lock et a. 1994). In 1999,2794 breedng pars of herrng gulls and 115 pars of great black-backed gulls nested on the entre sland (G. J. Robertson, unpubl. data). 1 conducted my research on Gull Island from 24 May to 15 August n 1998, and 16 May to 9 Augustn Four west-facng kttwake nestng clffs, whch were at least 200 m apart, were chosen as study plots (Appendx 1). To mnmze the dsturbance to breedng brds, all four plots were located at the southem end of the sland. The clff heghts of plots ranged from about 5-25 m. Three of the four dfts (N4, S5, SI) were wthn protected guches and one (PZ) was an open clff at the edge of Melane Masssmo - Ltuge gull pdaton on black-kgged ktttwlps 16

36 the sland. Indvdual study plots supporkd actve kttwake nests (2 1 egg was lad) Redaton behavout frequency Durng 1998 and 1999, predaton behavour frequency was quantfed on four kttwake study plots durng 2-4 h watches throughout the breedng seasm. Although the selecton of watches at study plots was not done randomly, 1 dstrbuted watches equdy, tempordy and spatally, among plots. AU observatons were done from blnds to ensure normal undsturbed predatory behavour of gulls. 1 also entered blnds approxmately 5 mn before a watch started to allow gulls to settle down after they were dstwbed by my &val. There was no evdence that predaton attempts were more frequent at the begnnng of a watch because of my approach to the colony. 1 defned a predaton attempt as an occason when a large gull closely approached, ether n flght or on foot, one or more kttwake nests, elctng responses such as turnng towards the gull and smutaneously loud callng, bu jabbng, peckng, btng and dvng at a gul on the clff ledge as well as durng flght n close proxmty to the clff. For each predaton attempt 1 recordec, whether a herrng gul1 or black-backed gull was nvolved. For each observaton perod, hourly predaton behavour frequency was caldated by dvdng the number of Melane Masssuro - Large gu1 prràaton on black-legged Inta<wkrs 17

37 predaton attempts by the nuxnber of observaton hours. Total observaton tmes were 286 h n 1998, and 426 h n obtaned predaton attempt rates for a total of 235 observaton perods, of whch 34 were drectly foowed by another watch at the same ste. On 21 occasons, two watches were done at the same locaton wthn one day, but several hours apart. I categorzed all watches nto four dfferent the-perods: early mornng ( ), mornng ( ), aftemoon ( ) and evenng ( ). I dsthgushed three perods wthn each breedng season: (1) from when 1 startec my research on the sland early n the season unal the mean date of herrng gull hatchng, (2) from the mean date of guü hatchng untl capeln amval, and (3) after capeln amval Breedng success In both years, all kttwake nests at the four study plots were numbered and mapped. Nest contents at al four plots were montored approxmately twce a week, except for P2 where no breedng data were coîlected n In montored a total of 700 kttwake nests and n 1999,645 nests. If kttwake chcks hatched between two watches, the date mdway between the two watches was taken as the date of hatchng measured to the halfday. Kttwake chcks that survved 35 days or more were consdered as fledged. In some rare

38 occasons kttwake chcks between the age of 30 and 35 days dsappeared or 1 left Gd Island too early to montor ther fledgng. Those ch& were assumed to fledge and were ncluded as fledged chcks n the analyss. If kttwake eggs or ch& were lost between two watches, the date mdway between the two watches was taken as the date of dsappeaance measured to the halfday. I classfed all kttwake eggs and chcks that were mssng between two nest checks as 'dsappeared'. If eggs broke, dead chcks were seen n the nest, or complete nests were mssng after heavy rans, 1 classfed those as egg or ch& loss. I calculated the percentage of eggs and chcks that dsappeared for each week of the year. 1 tested whether ths dsappearance rate was postvely correlated wth weekly mean predaton attempt rates. To obtan an estmate of the percentage of kttwake offsprng lost to gull predaton, 1 added the number of kttwake eggs and ch& that were seen to be taken by gulls to the number of offsprng that dsappeared, and dvded ths number by the number of eggs lad Tmng of guî hatchng In 1999,50 herrng gu nests, dstrbuted over the southem part of Gd Island, were randomly chosen and kked every three days untl all eggs had hatched. If chcks hakhed between two checksf the day mdway between the checks was Melune Masssmo - h ge gull predaton un buck-kgged kttuakes 19

39 consdered to be the hatchng date measured to the halfday. From a total of 46 nests wth 101 hatched eggs, the mean herrng gull hatch date was caldated. Unfortunately, the exact mean date of hatchng for herrng guls was not detennned for However, by 21 June most chcks were hatched (M. Massaro, pers. observaton) and ths date was used to defne the begnnng of herrng gull chck rea~g n that year. The sarne method was used for determnng the hatchng dates of great black-backed gulls (1998: n = 8 nests wth 18 hatched eggs; 1999: n = 10 nests wth 26 hatched eggs). Al1 observed black-backed guu nests were located at the southem end of Gu11 Island and were from soltary breeders, that nested wth a mnmum dstance of 20 m to the nearest ntra-speces neghbour Capeln arrval The date of frst delvery of capeln by Atlantc puffns and common mu- to ther chcks was taken as the date for nshore capeln arrval. In both years an abrupt ncrease n hurnpback whale (Mgctptera novaeanglae) numbers n Wtless Bay was observed smultaneously wth the frst delvery of capeln by breedng auks. Furthermore, n both yem the date of capeln arrval was confrmed by other observers, n 1998 by S. Balle (unpubl. data), who regularly cokcted puffn chck det data and n 1999 by an undemater flmaew, who dove

40 regularly close to G d Island. 1 used the terms 'nshore arrval' and 'frst spawnng arrval' nterchangeably throughout ths thess Statstcal analyses Gven my watches were not randornzed, watches mght not be completely ndependent. 1 was partcdarly concerned about the ndependence of back-toback watches. In order to test whether 1 could nclude d34 pars of watches done back-to-back n subsequent analyses, 1 tested whether the nurnber of herrng and great black-backed gull predaton attempts per hour occurrng durng the second watch of each par was ndependent of the number per hour occmng n 'he frst watch. 1 compared those 34 pars of watches wth the 21 pars of observaton perods whch were done at one plot wthn a day, but several hours apart. 1 caculated the dfferences n predaton attempt rates between the pars of watches and obtaned the rato of varances. Ths F-rato of a two-taled test dowed me to test whether there was a statstcdy sgnhcant dfference between predaton attempt rates obtaned from watches done backto-back, and watches done severa hours apart (Sokal and Rohlf 1995). Predaton rates were generally dstrbuted as a Posson dstrbuton, based on graphcal examnaton of the data and varance to mean ratos that approached 1.0. A generalzed ünear mode1 wth a Posson dstrbuted response Melane Masssmo - &ge gull prrdutm un back-kgged httwakes

41 varable (PROC GENMOD; SAS Insttute 19%) was used to compare predaton attempt rates. 1 ncluded the followng terms n the orgnal model: study plot, year, ntra-seasonal perod, tme of day and al1 two-way nteracton term. If statstcally > 0.1), hgh order terns were excluded from subsequent models untl only sgnhcant tem remaned. To reduce the rsk of a type II error 1 used a p < 0.1 to allow nteracton tems to reman n the model. However, for the fnal mode1 the tolerance for type 1 enor was set at 0.05 for man effects. Due to my samplng unt (number of gull predaton atternpts per hour) t was mpossble to nclude an ndependent varable for gull speces nto the man analyss (see above). To be able to compare the predaton attempt rates of the two gull speces ndrectly 1 chose to use herrng and great black-backed gull predaton attempt rates each as a response varable n two separate analyses. As n the man analyss I used a generazed lnear model wth a Posson dstrbuted response varable, ncludng the same four ndependent varables. 1 followed the sarne procedure as descxbed above for fndng the best fttng model. 1 used Pearson product-moment correlatons to test whether percentages of kttwake eggs and chcks that dsappeared were correlated (1) wth avaable eggs and chcks or (2) wth weekly mean observed predaton attempt rates. Melane Mnsssmo - Lmge grll predatun un bldr-kgged kttwakes 22

42 Except durng the process of fndng the best models, the tolerance for type 1 error was set at 0.05 for all other statstcal tests. AU tests were two-taled and al1 means are reported wth f 1 SD Results Tmng In 1998, schools of spawnng capeln frst arrved and spawned n Wtless Bay on 5 July; n 1999 capeln arrved nshore on 26 June, 9 days earler than n Mean hatchng dates for great black-backed gulls were 6 June ( 4 d) n 1998, and 2 June (f 7 d) n Mean hatch date of herrng gulls occurred on 9 June 1999 (& 5 d). Mean frst-egg layng dates for kttwakes were 2 June n 1998 (k 7 d) and 3 June n 1999 (f 12 d). Mean hatchng dates were 27 June n 1998 (f 6 d) and two days later n 1999 (29 June f 8 d). Medan layng and hatchng dates of gulls and kttwakes never dffered by more than one day from mean dates (Fg. 2.1). ne perod before mean herrng gull hatchng lasted from 4-20 June (17 d) n 1998, and from 18 May- 9 June (23 d) n The second perod, whch started after mean herrng gu hatchng and contnued untl capen arrva,

43 was 21 June- 4 July n 1998 (14 d) and June n 1999 (16 d). The perod after capeln arrva1 started on 5 July n 1998 and 26 June n In 1998, the last watch of the thrd perod was done on 7 Aug. (34 d), and n 1999 on 25 July (29 d; Fg. 2.1).

44 Perod 1 (174 Perod 2, I ( Perod 3 (344..._m_UNI I I HERG I 1 BLKI I, 1 I 1 Capeln Tme (weeks) I 1 1 b Perod 1 (234 Perod ;(164 I I 1 1 Perod 3 (294 GBBG."...U.-...I..."... 1 a I I Capeln Fg Tmng of breedng of great black-backed gulls (GBBG), h h g gulls (HERG) and black-legged kttwakes (BLKI) on Gu11 Island and tmng of capeln arrva1 n Wtess Bay n 1998 and Dotted lnes ndcate the perod when brds ncubated eggs and sold lnes descrbe when brds had chcks. Mean hatchng dates for al tluee brd speces and mean frst-egg layng dates for kttwakes are ndcated by bg crcles. The three ntra-seasonal perods for each year are reported wth ther duraton tme (d = days). Melane Masssmo - Lagc pl1 pdpton un bladr-kgged kffwpkes 25

45 Gd feedng terrtores Durng the two years of my study 1 observec that all kttwake plots were part of feedng temtores of breedng herrng and great black-backed gulls. Each kttwake plot was defended by one or - n case of 55 - two breedng, resdent gull pars aganst other ntrudng gulls. Study plots P2, N4 and S5 were each part of a feedng temtory of a dfferent resdent great black-backed gull par and SI and S5 were each occuped by one breedng par of h d g feedng temtores of three great black-backed gu pars and two These gull pars were consstent over both years Frequency of gull predatory behavour on kttwakes There was no dfference between predaton attempt rates obtaned from watches done back-to-back, and watches done severa hours apart on the same plot wthn a day (F al = 1.14, p > 0.05). Ths result aowed me to ndude al1 34 pars of watches done back-to-back nto the man andyss comparng gull predaton attemp t rates. Mean gull predaton attempt rates (attempts/hour) of hemng and great black-backed gulls on kttwakes at the four study plots are presented n Fgure 2.2. Whereas n 1998, mean predaton attempt rate per hou n the second perod (ncludng all plots) was about 6-7 hes hgher (1.72 f 1.6) than More Melane Mnsssaro - Lmge gull pdaton on blndr-egged b'ttrvakes 26

46 gu hatchng (0.27 k 0.34) and after capeln arrva1 (0.23 f 0.55), mean attempt rate n 1999 was only about 2-5 thes hgher (0.90 f 0-97) n the second perod than n the frst (0.41 f 0.66) and thrd perod (0.19 f 0.43). The fnal generalzed lnear modd for gull predaton attempt rates ncluded all man effects and one nteracton term, year by nha-seasonal perod, as predctors (Table 2.1). Gull predaton attempt rates dffered sgrufcantly among study plots, however there was no evdence of varaton arnong years. Predaton attempt rates were sgnfcantly dfferent among the three ntra-seasonal perods, wth hghest attempt rates n the second perod after mean henng gu1 hatchng, but before capeln arrva1 (Table 2.1). Although, mean predaton attempt rates were hghest early n the mornng (0.76 f 1.15), followed by attempt rates n the evenng (0.59 f 1.13) and lowest n the mornng (0.46 f 0.83) and aftemoon hours (0.45 f OH), those cfferences of attempt rates at dfferent tmes of the day were not statstcal sgnhcant (Table 2.1). The nteracton of year and perod was aso nsgmfcant n the fnal model (Table 2.1). For al1 analyses wth each gull speces treated separately, a two-way nteracton terms proved to be ether nspfcant or nsuffcent data were avalable to estmate the nteracton. There was a sgnhcant year and ntra- seasonal perod effect n h d g guü predaton attempt rates (Year ~ 2 9.7, = df Mehe Mnçssmo - Lmge gull pdaton on biack-legged kttwakes 27

47 = 1, p = ; Perod: ~ , - df = 2, p < ). However, for great black- backed gu predaton attempt rates, 1 found sgrufcant plot, ntra-seasonal perod and tme of day effects, but no sgmfcant year effect (Plot ~ , - df = 3, p < ; Perod: ~ , = df = 2, p < ; The of day: ~ 2 8.6, = df = 3, p = ). Mean predaton attempt rates of h-g gulls on kttwakes were lower durng al three perods n 1999 than n 199û. However, great black- backed gull predaton atternpt rates on kttwakes were smar between years (Fg. 2.3). By comparng the number of attempb made by each gull speces n each ntra-seasonal perod 1 found that most predaton attempts n the frst and thrd perod were made by great black-backed g ds (70.3% of ail attempts n the frst perod and 57.6% n the thrd perod). However, dutng the second perod 55.3% of al predaton attempts were performed by henng gulls and only 44.7% by great black-backed gds. Melane Mnsssmo - Lmge gull pàaton on &fa&-kgged kttwokes 28

48 Table 2.1. Generalzed lnear model of factors nfluencng rates of herrng and great black-backed guli predaton attempts on kttwakes on Gd Island n 1998 and Source df X' P PLOT 3 YEAR 1 PERIOD 2 TIME OF DAY 3 PERIOD * YEAR Note: Hgher order tenns not present were > 0.1) and dropped from the model. For the fnal model the tolerance for type 1 error was set a t 0.05 for man effects. Generalzed lnear mode1 wth Posson dstrbuton and a log lnk hncton. Melane Masssaro - Large gull pmhfun on bu-kgged k-s 29

49 Inta-seasonal perods Fg. 22. Mean gull predaton attempt rates (attempts/hour) on black-legged kttwakes at four study plots on G d Island wthn three ntra-seasonal perods of 1998 and The frst ntra-seasonal perod (1) lasted from the begnnng of the season untl mean h-g gull hatchng, the second perod (2) from mean h-g gu1 haahng untl capeln arrval, and the thrd perod (3) from capen arrva1 unt the end of the season. Melane Masssmo - Loge gu1 pre&ztan on back-legged ktt~ases 30

50 GBBG HERG Intra-seasonal perods Fg. 23. Mean predaton attempt rates (attempts/ hour) of great black-backed gulls (GBBG) and herrng on black-legged kttwakes on G d Island wthn three ntra-seasonal perods of 1998 and The frst nba- seasonal perod (1) lasted from the begnnng of the season untl mean herrhg gull hatîhng, the second perod (2) from mean herrng gull hakhng untl capeln arrval, and the thrd perod (3) from capeln arrval untl the end of the season.

51 24.4. Gu11 predaton attempt rates and chappearame of kttwake eggs and ch& 1 dd not fnd any correlaton between the percentage of kttwake eggs and chcks that dsappeared and avalable eggs and ch& n each week (r = 0.07, n = 21, p > 0.05). However, n both years, weekly gull predaton atkmpt rates were postvely correlated wth the percentage of kttwake eggs and chcks that dsappeared wthn each week (r = 0.77, n = 21, p < 0.05; Fg. 2.4).

52 Melane Masssmo - Lmge gull predaton on back-kgged kttwakes 33

53 Capeln d val M~Y Jtun J ~ Y Weeks of the year

54 2.45. The effect of gul predaton on kttwake bmdng success Of 700 observed kttwake nests n 1998,589 nests were actve (21 egg was lad). Of 1026 kttwake eggs lad, 686 chcks hatched and 230 eggs dsappeared. Of d chcks hatched, 367 fledged and 193 kttwake ch& dsappeared. In 1998, chck survval rate per nest was 0.62 (number of fledged chcks/number of actve nests) and 302 (51.3%) kttwake pars that lad eggs fledged ch&. An average of 0.52 chcks fledged per completed nest (Table 2.2). A total of 423 (41.2%) kttwake offsprng dsappeared and addng the 8 eggs and 8 chcks seen to be taken by herrng and great black-backed gulls, 42.8% (23.2% of al eggs and 29.3% of all chcks) of all kttwake offsprng were lost due to gull predaton n In 1999,515 nests were actve of 645 montored nests. Of 891 eggs lad, 657 chcks hatched and 158 eggs dsappeared. Of 657 chcks hatched, 480 fledged and 94 dsappeared. QUdc survva rate per nest was 0.93 and 329 (63.9%) pars wth eggs successful fledged chcks. An average of 0.72 chcks fledged successfdy per completed nest (Table 22). A total of 252 (28.3%) kttwake offsprng dsappeared and by addng the 13 eggs and 5 chcks that were depredated by herrng and great black-backed gulls, 303% (19.2% of dl eggs and 15.1 % of all chcks) of al1 kttwake offsprng were taken by gds n 1999.

55 Table 2.2 Breedng success of black-legged kttwakes on G d Island n 1998 and Kttwake breedng performance on Gull Island Year Total number of observa nests Total number of actve nests (2 1 egg was lad) Number of eggs lad Nurnber of eggs that dsappeared Number of chcks hatched Number of chcks that dsappeared Number of chcks fledged Chck survval rate (nurnber of fledged chcks/ number of actve nests) Average number of chcks fedged per completed nest

56 25. Dscusson Intra-seasonal varaton n gull predrtory behavour In both years 1 observed strkng varaton n the frequency of gull predatory behavour tluoughout the breedng season. At al plots predaton attempt rates ncreased after mean hehg gull haahng and decreased a h capeh amval. What caused those drastc dfferences n gull predaton attempt rates over the breedng season? Threlfall(1%8) noted that herrng gulls fed manly on blue mussel (Myflus eduls) n Wtless Bay dumg May and June and changed to capeln as a major food source later n the season. Perott and AMett (1987) ewmned the det and tmng of prey-swtdng by h&g guls n Wtless Bay from 1976 to In those years capeln arrvec n Wtless Bay early n June (Perott and Annett 1987). They found that herrng gulls swtched to capeln as soon as they had chcks to feed, rather than when capeln, the most proftable prey tem, becarne avalable. Snce the begnnng of the 1990s, delayed capeln arrval and the groundfsh moratorum have substantdy decreased food avalablty at a cmcal tme when gulls have small chcks to feed. Low guî breedng success n Wtless Bay (Neuman 1994; Regehr and Montevecch 1997; J. W. Chardne, unpubl. data) and on the north shore of the Gulf of St. Lawrence (Chapdelane

57 and Ral 1997) snce 1990 support the dea that large gulls suffered from low food avdablty and have been forced to fnd alternatve food resources, such as kttwakes, puffns and lead's stom-petrels. Stenhouse and Monkvecch (1999) found that herrng gull predaton on adult leach's storm-petrels decreased markedly after capeh arrval. Russell and Montevecch (19%) suggested that ktawake offsprng and adults appear to be easer targets for guls than adult puffns. Addtonally the offsprng of kttwakes are vsually 'avalable' durng the perod of hgh food demand compared to puffn and the nocturnal leach's stom-petrel offsprng whch are protected by burrows. In contrast to the stuaton pror to 1990, a lack of capeln after gull chcks hatched appears to have caused an ncrease of gull predaton attempts on kttwakes. However, as soon as spawnng capeh became avalable to chck-rearng gus, they once agan foraged on capeln, whch offers a low-rsk, hgh-energy food resource (Perott and Annett 1987). Low herrng and great bladc-backed gui breedng success mght also be a result of h& rates of cannbalsm n both speces. It would be to great nterest to nvestgate the levels of ca~balsm pnor and pst capeln asrval.

58 2-5-2 Inter-year cfferences n gull predaton rates Although no sgrufcant dfference n predaton attempt rates between years was detected n the man analyss, 1 observed hgher mean gull predaton attempt rates n d three perods n 1998 cornpared to 1999, except durng the frst perod at P2 and the t.d perod at N4. Ths suggests that n 1999 the overd food avalablty was hgher than n Earler capeln arrval, a longer perod of capeln avalablty (M. Massaxo, pers. observaton) and a 53.3% ncrease n kttwake breedng success supports ths suggeston. Although spawnng capeln arrved n Wtless Bay 9 days later n 1998 ttw n 1999,I dd not observe a longer perod between mean gull hatchng and capeh arrval n Due to the fact that the exact date of herrng g d hatdng n 1998 was unknown, the perod between g d hatchng and capeln arrva mght have been longer than assumec n ths study (>14 d). However, based on casual observatons, 1 doubt that mean h d g gull hatchng occurred sgnucantly earler than 20 June. The nter-year dfference of rnean gui predaton attempt rates s caused by the drastc varaton of hermg gull predaton rates among years. Howevet, great black-backed gull predaton attempt rates on kttwakes were smar between years. Why dd predaton rates by herrng guh, but not by bla&- backed gulls, dffer arnong years? n both years, three of the kttwake study Melanr Mnsssmo - Lwge gull pdaton a back-egged kttulakes 39

59 plots were dstnct feedng terrtores of great black-backed g d pars. Spea (1993) observed that male western gus (Lmus occdentals), whch had specalzed n feedng on cornmon murres and brandt's cornorants (Phnlacrocoru pencllahs), showed hgh fdelty to ther feedng temtores. In 1999 on Gul Island, 1 unquely colour banded one male great black-backed gull, whch held a feedng terrtory at kttwake study plot N4. Skty percent of dl great black-backed gull predaton attempts on kttwakes at N4 were made by ths ndvdual (M. Massaro, unpubl. data). Whle great black-backed gulls were responsble for most predaton attempts on kttwakes durng the frst and thrd perod, herrng gulls pursued more predaton attempts durng the second perod. On only one occason was a great bladc-backed gull observed to ntrude on the feedng terrtory of another great back-backed gull. However, on 34 occasons, we observed terrtoral defense behavour n whch herrng gulls were the ntruders. Usually durng such gu-gull nteractons the resdent guli approached rapdly another gu and forcng t to leave the temtory by chasng and attemptng to bte the ntrudng guu. Twenty-fve (74.3%) of al ntmsons were observed durng the perod between gull hatchhg and capeln arrva1 (M. Massaro, unpubl. data). Ths suggests that for most herrng gulls pursung a predaton attempt n a black-badced gull terrtory s a rs4 venture, whch should be avoded f other food sources are avalable.

60 Zm5.3m Gd predaton rate dfferences among study plots Includng both years, n a three perods P2, the open and more exposed dff had the hghest mean predaton attempt rates. P2 was followed by N4 and lowest mean predaton attempt rates were obsewed at SI. At SI, only h d g gulls were observed as kttwake predators. SI was a very small clff whch could not support much more than 50 kttwake nests. The low number of kttwake breedng pars at SI, and the fact that no resdent great black-backed gull par occuped S1 as a feedng temtory, mght explan the low predaton attempt rate compared to the other study plots. Although Regehr et al. (1998) stated that large-scale clff structure nfluences the predatory behavour and success of avan predators on kttwakes, causng cfferences n kttwake breedng performance, they dd not observe and compare gull predaton rates among clffs. Dfferences n kttwake breedng performance among clffs mght be a result of a varety of factors nteractng wth each other, such as exposure of the clff to wnd, the nurnber of resdent predatory gulls, ectoparaste abundance, qualty of breedng brd and large-scale clff structure. Although P2 had the same number of resdent predatory pls as other clffs (N4, S5), predaton attempt rates were hgher. 1 beleve that due to the open dff structure, wnd condtons at P2 on most days were more favorable for gu1

61 predaton than n narrow gulches, where wnds are gusty and gulls cannot maneuver as easly (Gkhrst and Gaston 1997; Glchrst et al. 1998) ne mpact of gull predaton on cttwake bnedng performance At the populaton level t s essentd to know the mpact of large gull predaton on kttwakes and ts mplcatons on kttwake breedng performance. The demonstrated relatonshp between gull predaton attempt rates and the number of kttwake offsptng that dsappeared supports the assumpton that most mssng kttwake eggs and chcks were lost due to gui predaton. Durng two seasons of ntensve observatons of kttwake colones only one egg and four dcks were seen to fall out of a nest n the absence of a predaton attempt. Kttwake reproductve success n Newfoundand has been Iow çnce at least 1990 wth the excepton of 19% (Neuman 1994; J. W. Chardne, unpubl. data). At the Ga.n.net Islands n Labrador, Canada, kttwake breedng success ranged from zero to 0.77 fledged ch& per nest n , compared to hgher success n rangng from 0.90 to 1.l3(Hpfner et al. 2000). A sunlar decrease n kttwake breedng success has been observed n southeastem ScotIand and northeastern England snce 1986, explanec by largescale ndustral fsheres for sandeds (Anzmoàyfes marnus), a major prey tem for kttwakes n the North Sea (Harrs and Waness 1990; 1997). It has been

62 shown that changes n populatons of marne prey speces have dted mpacts on seabrd breedng success (e.g. Vermeer et al. 1979; Bard 1990; Hamer et al. 1993; Barrett and Krasnov 1996; Harrs and Wanless 1997), however, ndrect effects, due to ncreasng predaton of predatory brd speces on other brds, have been rarely studed (Hamer et al. 1991; Spear 1993; Stenhouse and Montevecch 1999). In years of low food avalablty n Wtless Bay, kttwakes are confronted not only wth dffdtes n provdng chcks wth food but also wth an ncreased predaton pressure by gulls. Addtonally, n years of food shortage kttwakes show low adult attendance at nests wth chcks, and the rsk of nestlngs beng depredated ncreases (Barrett and Runde 1980). On Great Island n Wtless Bay ths caused a complete breedng falure of kttwakes n 1992 (1% of pars wth eggs fledged chcks; Regehr 1994). In 1998 and 1999 on Gull Island chck survval was lower than n most other Atlantc studes [53.5% (1 998) and 73.1 % (1 999); 88 % Cullen (1957); 87 % Coulson and Whte (1958); 81 % n 1969 and 73% n 1970 Maunder and Threlfal(1972); 56% n 1973 and 75% n 1974 Barrett and Runde (19ûû); 68% n 1993 Regehr and Montevecch (1997)], but hgher than n some studes [26% n 1976 Barrett and Runde (1980); 7% n 1992 Regehr and Montevecch (IgW)]. On Gull Island 0.52 (lm) and 0.72 (1999) chcks fledged per completed nest. Harrs and Wanless (1990) reportec rates for young fledged per completed nests between zero and 1.56 for 36 kttwake

63 colones at the coast of England, Scotland and Ireland. In kttwake colones n France, Danchn and Monnat (1992) ohed 1.03 fïedged young per pav n a flourshng colony compared to only 0.49 young per par n a declnng colony. My estmates are below 1.03 and closer to Changes n abundance of marne prey speces, caused by fshng actvtes or clmate change (Stergou 1991)' have ndrect effects on the predaton pressure rnposed by large gulls on other seabrds. The predaton of western gulls on common murres and brandt's cornorants were sgntcantly hgher durng an El No year compared to the other years (Spear 1993). Gulls took 66% of d eggs lad durng the El NTo year compared to 18% and 12% durng years of normal food avdablty. Compared to Spear's study my estrnates of the overall mpact of gu1 predaton on kttwakes are hgh, consderng that my estmates are markedly hgher than 18% n two subsequent years (1998: 42.8%; 1999: 30.3%). Regehr's (1994) study suggests that the mpact of gull predaton on kttwakes was even hgher n the early 19% than n 199û and Hgh rates of large gull predaton for almost a decade now may have an mpact on the growth rate of kttwake populatons n Newfoundland. However, the relatvely good season for kttwakes n 1999 suggests that the overall food stuaton s mprovng and decreasng gull predaton rates on k ttwakes. Melme Mnsssmo - Lage gull pre&ton on black-leggcd kttgarjrcs 44

64 2.6. References Bard PH (1990) Influence of abotc factors and prey dstrbuton on det and reproductve success of three seabrd speces n Alaska. Oms Scand 21: Barrett RT, Krasnov W (19%) Recent responses to changes n stocks of prey speces by seabrds breedng n the southem Barents Sea. ICES J Mar Sc 53: Barrett RT, Runde OJ (1980) Growth and survval of nestlng Kttwakes Rssu frùzctya n Norway. Oms Scand 11: Beaman MAS (1978) The feedng and populaton ecology of the Great Black- backed G d n northern Scotland. Ibs 120: Brown RGB, Nettleshp DN (1984) Capeln and seabrds n the northwest Atlantc. In: Nettleshp DN, Sanger GA, Sp~ger PF (eds) Marne brds: ther feedng ecology and commercal fsheres relatonshps. Specal Publcaton, Canadan Wldlfe Servce, Ottawa, Canada, p Burger AE, Patt JF (1990) Flexble tme budgets n breedng common murres: buffers aganst varable prey abundance. Stud Avan B01 14: 7l%3 Burger J, Gochfeld M (1984) Great black-backed gull predaton on kttwake fledglngs n Norway. Brd Study 31:

65 Carscadden JE, Nakashma BS, Frank KT (1997) Effects of fsh length and temperature on the tmng of peak spawnng n capeln (Mnlbhs vllosus). Can J Fsh Aquat Sc 54: Chapdelane G, Ral J-F (1997) Relatonshp between cod fshery actvtes and the populaton of herrng gulls on the North Shore of the Gulf of St- Lawrence, Quebec, Canada. ICES J Mar Sc 54: 7Oû-7ï3 Coulson JC, Whte E (1958) Observatons on the breedng of the Kttwake. BW Shdy 5: Cullen E (1957) Adaptatons n the Kttwake to dff nestng. Ibs 99: Danchn É, Momat J-Y (1992) Populaton dynamcs modellng of two neghbourng kttwake Rssu hdactyk colones. Ardea 80: Furness RW, Monaghan P (1987) Seabrd Ecology. New York, Chapman & Hal Furness RW, Ensor K, Hudson AV (1992) The use of fshery waste by gu populatons around the Brtsh Mes. Ardea 80: Garthe S, Camphuysen KCJ, Fumess RW (1996) Arnounts of dscards by commercal fsheres and ther sgnhcance as food for seabrds h the North Sea. Mar Eco1 Prog Ser 136: 1-11 Glchrst HG, Gaston AJ (1997) Effects of mue nest ste characterstcs and wnd condtons on predaton by glaucous gulls. Can J Zoo1 75: Melane Masssmo - Large gull pndaton a black-legged kltmokes 46

66 Glchrst HG, Gaston AJ, Smth JNM (1998) Wnd and prey nest sks as foragng constrants on an avan predator, the glaucous gull. Ecology 79: Glchrs t HG (1 999) Declnng thck-blled mue Urn hus colones experence hgher gull predaton rates: an ntercolony cornparson. Bol Cons 87: Hamer KC, Fumess RW, Cadow RWG (1991) The effects of changes n food avalablty on the breedng ecology of great skuas Gzthmactn skun n Shetland. J Zoo1 (Lond) 223: Hamer KC, Monaghan P, Uttley JD, Walton P, Bums MD (1993) The nfluence of food supply on the breedng ecology of Kttwakes Rssa hdrrctyza n Shetland. Ibs 122: Harrs MP (1965) The food of some L Ibs 107:43-53 Harrs h4p (1970) Rates and causes of ncreases of some Brtsh gu populatons. Brd Study 17: Harrs MP (1980) Breedng performance of Puffns Fraterculn mctca n relaton to nest densty, layng date and year. Ibs 122: Harrs MP, Wanless S (1990) Breedng success of Brtsh kttwakes Rssa h&cfyîa n : evdence for chgng condtons n the northern North Sea. J Appl Eco1 27: Melme Masssmo - Large pl1 pre&fm on blpdr-legged kttha&es 47

67 Harrs MP, Wanless S (1997) Breedng success, det, and brood negleft n the kttwake (Rssa hdaftyla) over an Il-year perod. ICES J Mar Sc 54: Hpfner JM, Adams PA, Bryant R (2000) Breedng success of Black-legged Kttwakes, Rss br'dactyla, at a colony n Labrador durng a peod of low capeln, Mnlbtus vllosus, avalablty. Can Feld-Nat 114. In press Howes LA, Montevecch WA (1993) Populaton trends and nteractons among tems and gulls n Gros Morne Natonal Park, Newfoundland. Can J Zoo1 71: Hudson AV, Fumess RW (1989) The behavour of seabrds foragng at fshng boats around Shetland. Ibs 131: Hutchngs JA, Myea RA (1994) What can be learned from the colapse of a renewable resource? Atlantc Cod, Gadus mothun, of Newfoundland and Labrador. Can J Fsh Aquat Sc 51: Jangaard PM (1974) The capeln (Ahlotus ollusus): bology, dstrbutons, explotatons, utlzaton and composton. B d Fsh Res Bd Can 186: ~OPP Kadlec JA, Dnry WH (1968) Structure of the New England Herrng Gd populaton. Ecology 49:

68 Lock AR, Brown RGB, Gerrets SH (1994) Gazetter of marne brds n Atlantc Canada. Canadan Wdlfe Servce, Atlantc Regon, 137pp. Maunder JE, Threlfall W (1972) The breedng bology of the black-legged kttwake n Newfoundland. Auk 89: McAUster DE (1963) A revson of the Smelt famly Osmerîdae. Nat Mus Can Bull 191: 1-53 Myers RA, Cadgan NG (1995) Was an naease n nahral mortalty responsble for the coapse of northem cod? Can J Fsh Aquat Sc 52: Neuman JA (1994) Aspects of the behavour and ecology of black-legged kttwakes, Rssa hduc tyla, breedng at two stes n Newfoundland, Canada, MSc thess, Mernoral Unversty of Newfoundland, St. John's Newton 1 (1993) Predaton and lmtaton of brd numbers. Current Ornthology 11: Perott R, hnett C (1987) Reproductve consequences of detary specalzaton and swtchng n an ecologcal generalst. In: Kaml AC, Krebs JR, Pullam HR (eds) Foragng behavour. Plenum Ras, New York, p

69 Regehr HM (1994) Breedng performance of black-legged kttwakes on Great Island, Newfoundland, durng perods of reduced food avaüablty. MSc thess, Mernoral Unversty of Newfoundland, St. John's Regehr HM, Montevecch WA (1997) Interactve effects of food shortage and predaton on breedng falure of black-legged kttwakes: ndrect effects of fsheres actvtes and mplcatons for ndcator speces. Mar Eco1 Prog Ser 155: Regehr HM, Rodway MS, Montevecch WA (1998) Antpredator benefts of nestste selecton n Black-legged Kttwakes. Can J Zoo1 76: Regehr HM, Rodway MS (1999) Seabrd breedng performance durng two years of delayed capeln arrva n the northwest Atlantc: a multspeces cornparson. Waterbrds Russell JO, Montevecch WA (1996) Predaton on adult Puffns Frntercula mctfn by Great Black-backed Gds uns marnus at a Newfoundand colony. Ibs 138: SAS Insttute (19%) SAS system for Whdows, Reease SAS Insttute, Inc, Cary, NC Schauer JHS, Murphy EC (1996) Predaton on eggs and nestlngs of Common Murres (Utn aalge) at Bluff, Alaska. Col Waterbrds 19: 1S198 Melane Masssmo - Lngc gull p&on a bladr-legged h'ttmalps 50

70 Shackell N, Carscadden JE, Mller DS (1994) Mgraton of pre-spawnng cap* (Mallofus Mllosus) as dated to temperature on the northern Grand Bank, Newfoundland. ICES J Mar Sc SOM RR, Rohlf FJ (1995) Bometry, 3rd edton. Freeman, New York Spear L (1993) Dynamcs and effect of westem gulh feedng n a colony of gullemots and Brandt's cormorants. J Anm Ecol 62: Stenhouse 1, Montevecd WA (1999) Indrect effects of the avalablty of capeln and fshery dscards: gull predaton on breedng storm-petas. Mar Ecol Prog Ser 184: Stergou KI (1989) Capen, Mzllohts ullosus, glacatons and specaton: a nomothetc approach to fsheres ecology and reproductve bology. Mar Eco1 Prog Ser 56: Stergou KI (1991) Possble mplcatons of clmatc varablty on the presence of capeln (Mallotus vllosus) off the Nomegan Coast. Clmatc Change 19: Templeman W (1948) The Me hstory of the capeln (Mallotus vllosus O. F. Müller) n Newfoundland waters. Bul Newfoundland Gov Lab 17:l-151 Themault TW, Schneder DS, Methven DA (1996) The tmng of sparmng n capeln (Mdlohcs allosus Müller) at a coastd locaton n easten Newfoundland. Polar B01 16: Melune Masssaro - Large gull predatun on bnck-kgged kttwokes 51

71 Threfall W (1968) The food of three speces of gulls n Newfoundland. Can Feld Nat 82: Verbeek NAM (1979) Some aspects of the breedng bology and behavout of the great black-backed gd. Wlson Bull 91: Vermeer K, Cden L, Porter M (1979) A provsonal exphnaton of the reproductve falure of Tuh Puffns Lunda drrhnta on Trangle Island, Brtsh Columba. Ibs 121:

72 CHAPTER 3 Relatonshps between black-legged kttwake nest-ste characterstcs and susceptblty to predaton by large gulls 3.1. Abstract To understand how certan kttwake ste characterstcs, plot varablty and wnd condtons affect large gulls foragng ablty 1 quantfed the relatonshp between black-legged kttwake nest-ste characterstcs and rsk of predaton by great black-backed and hehg gulls at Gull Island, Wtless Bay, Newfoundland, Canada durng 1998 and montored kttwake nestng clffs to dentfy stes attacked by large gulls and compared characterstcs of depredated and survvng nests arnong four study plots. I dso examnec whch nest stes were attacked by herrng gulls and great black-backed gulls durhg carn (< 10 km/h) and wndy condtons (1 10 km/h). 1 found that ndvdual kttwake nests at the smallest plot were more lkely to be attacked by large gulls compared to nets on larger plots. Hence, the percentage of faled nests was hghest at the d e s t plot and nestng success naeased as the sze of the plots (number of nests) naeased. Nestng densty and the locaton on the dfl Melane Masssmo - hge gull prehtnt a black-le@ ktth&s 53

73 were nest-ste characterstcs that reduced sgnücantly the rsk of predaton by large gulls. Breedng success was correlated wth ledge wdth and nestng densty and dffered sgruhcandy among plots. Regardless of wnd condtons both gul speces attacked nest stes located on upper parts of clffs to a hgher percentage than nests located on mddle or lower parts of the clff. However, durng calm condtons, roofs over nest stes reduced the rsk of predaton by herrng gulls, whereas stes locakd on narrow ledges were less lkely to be attacked by great black-backed gus. Durng wndy condtons, nestng densty affected whch stes were attacked by great black-backed gulls. Taken togethet, rny results demonstrated a hgh rate of predaton of kttwake nests by large gds at Gd Island, wth strkng dfferences among plots. 1 also demonstrated that kttwake nes t locaton, certan nes t-ste characterstcs, and breedng densty can all rduence the rsk of predaton.

74 3.2 Introducton Long-lved speces that nvest heavly n rasng ther few offsprng are expected to select breedng habtat that maxmzes the chances of sumval for ther offsprng and themselves (Lack 1954; Martn 1988). Rsk of predaton s lkely to be one of the most mportant factors duencng habtat selecton (Martn 1993; Danchn et al. 1998; Rachlow and Bowyer 19!38). In brds, the selecton of clffs as breedng habtats lkely evolved as a response to predaton by terrestral mammals (Cullen 1957; Tuck 1%1; Brkhead et al. 1985). However, clff-nestng does not necessary protect brds from avan predaton. Several studes have shown that a number of speces n the famles Lardae and Corvdae are successful n preyng upon clff-nestng brd speces (Montevecch 1979; Maccarone 1992; Gaston and Ellott 19%; Glchrst and Gaston 1997; Barbraud 1999). Unlke most gulls, black-legged kttwakes and red-legged kttwakes (R. brevrosfns) breed on steep, vertcal clffs rather than on the ground. Caen (1957) proposed that as kttwakes evolved, they swtched from ground-nestng to clff-nestng to avod predaton by terrestral predators. Wth the adopton of clff-nestng, well-developed mobbng behavow as a group defense tactc aganst predators, such as observed n many speces of Lardae, was lost or at Melmc Mnsssmo - Lngc gull predntun on black-kgged ktüwakes 55

75 least much reduced n kttwakes (Cullen 1957; Shealer and Burger Cavanagh and Grffn 1993; Yoro and Quntana 1997). However, kttwakes may defend ther offsprng aganst predators f faced wth a hgh rsk of predaton. Whereas many kttwake colones, n partcular n Brtan and Alaska, lose few or no eggs and chcks to large gulls, ravens or crows (Coulson 1963; Maunder and Threlfall1972; Murphy et al. 1991), other kttwake colones experence greater rates of avan predaton (Barrett and Runde 1980; Maccarone 1992; Regehr and Montevecch 19%). On the southeastern coast of Newfoundland, Canada, ncreasng populatons of herrng and great blackbacked gulls and reduced avalablty of naturd marne food resources and fsheres waste snce the early 19%, have caused ncreased predaton by large gulk on kttwakes (Regehr and Montevecch 1996; Massaro et al. 2000). Under condtons of hgh predaton pressure, spectc net-ste characterstcs, nestng densty and colony sze rnay play a role n enhancng reproductve success. Nestng n large colones wth hgh nest denstes may offer advantages to ndvdual brds due to ncreased vgrlance, group defense and predator swampng (Burger and Gochfeld 1994, Wttenberger and Hunt 1995, Glchrst and Gaston 1997). In clff-nestng thck-blled murres, nest-ste characterstcs and breedng densty nnuenced foragng success of glaucous gulls preyng on murre eggs (Glchrstand Gaston 1997; Glchrst et al. 1998).

76 Kttwake nest-ste characterstcs may be partcularly mportant n reducng predaton late n the breedng season when large kttwake chcks are often left alone by ther parents, ncreasng ther vulnerablty to predaton. On Baccaleu Island, Newfoundland, Maccarone (1992) found that common ravens (Cwmcs corax) were more lkely to patrol dong upper parts of a kttwake clff ather than the mddle or Iower parts. Kttwake nests wth chcks had larger overhangs than randomly selected nests on Great Island, Newfoundand (Regehr et al. 1998). That study used survval tme of kttwake chcks and eggs to nf er w hc h nest-ste characters tcs may reduce gull preda ton. Kttwake colony sue, specfc nest-ste characterstcs, nestng densty and wnd condtons may constran the foragng success of large gds depredatng kttwakes. The objectve of ths study was to quanhfy the relatonshp of kttwake nest-ste characterstcs and susceptbüty to herrng and great black-backed gul predaton at Gull Island, Newfoundland. I montored nests to detennne whch stes were and were not attacked by large gds. 1 compared characterstcs of nest stes that were attacked wth stes that were not to test whether plot, ledge wdth, roof, number of walls, nestng densty or dff part affected whch stes were attacked. 1 also compared whch ne& were attacked by herrng gulls and great black-backed gulls durng calm and wndy Meme Masssaro - Lmge gull predaton an bacù-legged kttzuukes 57

77 condtons. Af ter denafyng whch nest-ste characterstcs reduced the rsk of gull predaton, 1 examned whether these same nest-ste characterstcs nfluenced breedng success Methods Study locaton The study was conducted from 24 May to 15 August 1998 and 16 May to 9 August 1999 on Gd Island (47' 16' N, 52' 46' W), the most northerly of four slands wthn the Wtless Bay Seabrd Ecologcal Reserve off the southeastem coas t of Newfoundland, Canada ( Appendx 1). Gd Island s ap proxmatel y 1.6 km long and 0.8 km wde. The cffs of Gd Island offer breedng habtat to over 10,000 pars of black-legged kttwakes, approxmately 175 pars of razorbls ( Ah tord@, and 7ûû pars of common mmes (Lock et al. 1994). The sland also supports approxmatey 2,800 breedng pars of herrng gulls and 115 pars of great black-backed gulls (G.J. Robertson, unpubl. data). Melane Masssaro - h ge gdl ptedatm un black-legged kttwakes 58

78 Study plots In order to examne the nauence of nest-ste characterstcs on predaton, 1 selected four west-facng kttwake clffs as study plots, whch dffered n sze, heght, and overall cff structure. AU four plots were located at the southem end of the sland, but were at lest 200 m apart (Appendx 1). Three of the four plots (N4, S5, SI) were wthn protected gulches (narrow nlets) and one (PZ) was part of an open clff face a the edge of the sland. For all plots wthn a gulch 1 determned the openng angle towards the sea by recordng the angles from the narrow end of a gulch dong both clffs enclosng Of all study plots, S1 was the smallest clff wth an approxmate heght of 5-6 m. The guch had an openng angle of approxmately 16". P2 was the second largest clff wth a heght of approxmately 8-9 m and was not located wthn a gulch. N4 was m Iugh and the openng angle towards the sea was approxmately 20'. Beng approxmately 20 m hgh, plot S5 was the hghest clff studed. The openng of the gulch was approxmately 19" Kttwake nest predaton In both years al nests at the four study plots were ndvdually numbered, mapped and photographed. At ali four kttwake plots, 24 h watches were regularly conducted throughout the breedng season. Total observaton tmes Melane Masssaro - Lmge gull predaton a black-egged kttwakes 59

79 were 286 h n 1998, and 426 h n 1999, of whch 192 h were at N4,159 h at S,l% h at P2, and 165 h at Sl. To ensure nomal, undsturbed predatory behavour of guus all observatons were made from blnds, and 1 entered blnds approxmately 5 mn before a wakh started to allow gulls to settle down after they were dsturbed by my arrval. There was no evdence that predaton attempts were more muent at the begnnng of a watch due to my approach to the colony. For each herrng gu1 or great bladc-backed gull predaton attempt, whether successful or unsuccessful, 1 recorded whch nest was attacked. If more than one kttwake nest was attacked durng a predaton attempt 1 chose one of the nests randomly to ensure ndependence and ncluded t n the analyses. Nests that were attacked by herrng or great black-backed gulls at Ieast once n 1998 or 1999 were classfed as 'attacked' nests Nest-ste characterstcs To mnmze dsturbance to breedng bùds, nest-ste characterstcs were quantfed durng late dck rearng by observaton from a dstance of about m wth bnoculars and a 30x75 spottug scope. For al kttwake nest stes at each plot 1 recorded fve characterstcs wth the followng categores: (1) ledge wdth (broad, narrow), (2) roof (roof, no roof), (3) number of vertcal walls (zero or one, two or three walls), (4) nesüng densty (low, or h&) and

80 (5) clff part (upper, mddle or lower). I dassfed the dfferent categores of ledge wdth, roof and walls folowng Gaston and Nettleshp (1981; Fgure n). A nest was consdered to be on a narrow ledge f the nest materal was hangng over the edge of the seaward-orented ledge. A nest ste had a rmf f the nteror of a nest was overhung by rock wthn twce the heght of an adult kttwake. 1 counted the number of walls rnmedately surroundng the nest. Walls had to be at east the heght of an adult kttwake sttng on a nest and were at leaçt as wde as a nest cup. For the varable 'nestng densty' I counted for each nest ste the number of drect neghbourng breedng pars wthn a radus of three body lengths of a standng kttwake (approxmately a radus of 0.75 m; n * r* = 1.77 m2) and added t to one (for the resdent kttwake par on each ste). All stes wth a nestng densty ndex (1) of one or two were classfed as low densty ( kttwake pars/m2), (2) of three were dassfed as densty (1.7 kttwake pars/m2), and (3) of four and more were consdered stes wthn hgh densty areas (12.3 kttwake pars/m2). To determne whether a ste was located on the upper, mddle or lower part of the clff 1 quantfed the dstance between each nest ste and the uppa edge of the clff. I used a Bushnell laser range-fnder to measure the dstance between the observer and (1) the nest-ste, (2) the upper edge of the cff, and (3) the pont on the clff whch was at eye-level, to the nearest meter. A dnometer was used to

81 measure the angles between those three ponts, and the dstance of a nest to the upper edge of a dff was calculated tgonometrcally. For each kttwake plot 1 calculated the meda. dstance between all ne& and the cws' upper edge. I defned the border between upper and rnddle clff by subtractng one-half of the standard dwaton from the medan. By addng half a standard devaton to the medan the border between the rnddle and lower part of the clff was detemned (Table 3.1). Ths procedure was used to defne upper, mddle and lower part of all four study plots. The number of nests and the percentage of nestç wth a certan nest-ste characterstc are reported n Table 3.2 for each study plot. Nest-ste characterstcs for each ndvdual nest ste are lsted n Appendx 2.

82 Table 3.1. Defnton of the upper, mddle and lower part for four study plots (SI, P2, N4 and S5) on Gull Island n 1998 and The dstance of a kttwake nest to the upper edge of a clff defned where a nest was located relatve to the upper edge of a dff. Hot SI Plot P2 Plot N4 Plot S5 Medan = Medan = Medan = Medan = 1.70 m 2.16 rn 4.86 m 4.85 rn (I 1.03) (I 1.90) (I 2.25) (k 4.09) UPPER PART < 1.19 m < 1.21 m < 3.74 m < 2.81 m MIDDLEPART m m m m LOWER PART > 2.21 m > 3.11 m > 5.98 m > 6.89 m Note: Medan nest dstance to the dff's upper edge and f 1 SD are recorded for each plot.

83 Table 3.2. The number of actve kttwake nests C> 1 egg was lad) fallng wthn each level of nest-ste characterstc, for each of the four study plots on Gull Island n 1998 and Nest-ste characterstcs Pot SI, Plot P2, Plot N4, Plot S5, n=42 nt81 n = 268 n= 268 LEDGE WIDTH ROOF broad narrow roof no roof NO.WALLS Oorlwall DENSm low hgh CLEF PART upper part rnddle part lower part Melane Mosssmo - Lmge gull preàatm on black-kgged kttzuakes 64

84 Wmd condtons Durng both seasons, wnd speed and drecton were measured houry by a weather staton (Davs nstruments Corp., Weather Wzzard III) located on an exposed hll at the southwestern end of Gd Island (Appendu 1). The measurng devce was fxed to a tree trunk approxmately 1.8 m above ground. Due to a programmng mstake the weather staton dd not collect any data for May and June To compensate, I measured wnd condtons wth a handheld anemometer durng most predaton behavour watches at the kttwake study plots. For 16 occasons the wnd condtons durng gull predaton attempts were unknown and 1 used wùd data colected by Envronment Canada at the St. John's arport, located approxmately 35 km to the north of the study area. For each gull predaton attempt 1 classfa the wnd condton ether as calm 10 h/h) or wndy (> 10 km/h). However, f wnd drectons were not wthn the openng angles of the clffs, wnd condtons were consdered to be calrn even f wnd speed exceeded 10 km/h. For PZ, only westerly (180' - 360') wnds over 10 km/h speed were consdered "wndy". For al other three plots only south-south-easterly to south-south-westerly wnds (157.5' ") over 10 km/ h were consdered "wndy".

85 Breedng succesa of kttwakes Contents of all nurnbered nests at the four plots were montored appmxmately twce per week, except at plot P2 where no breedng data were colected n For events that occurred between observaton perods, such as chck hatchng, the date mdway between the two watches was taken as the date of the event, measwd to the nearest haü day. Kttwake chh that survvec 35 days or more were consdered to have fledged. On rare occasons kttwake chcks between the age of 30 and 35 days dsappeared (II = 76 chcks); those chcks were assumed to have fedged and were nduded n the analyss (9.0% of al chcks fledged). AU nests where one or more chcks fedged successfuy n 1998 or 1999 were consdered 'successfu' Statstcal analyses Only kttwake nests that were actve (51 egg was lad) n 1998 or 1999 were nduded n analyses. If a nest ste was usa by kttwakes n 1998 and 1999, the nest ste was only counted once to avod pseudo replcaton. Smlady, f a kttwake nest-ste was attacked by gulls n 1998 and 1999, t was only nduded once n the dataset To test whether the proporton of attacked and successfd nests dffered among plots, 1 wd ch-square tests.

86 To be able to nclude densty as a ndependent varable n the man analyss 1 chose randomly 300 stes (out of 659 actve stes) and used these as my sample sze. Ths procedure ensured ndependent samples. To test whether nest-ste characterstcs nfluenced susceptblty to large gull predaton 1 used a generalzed lnear model wth a bnary respohse varable (attacked or not attacked), a logt lnk functon and sx dme te ndependent varables: plot, ledge wdth, roof, number of walls, densty and clff part. All two-way nteractons were ncluded n the orgnal model and excluded from the fnal model f the probablty was hgher than 0.1 n the orgnal model. To reduce the rsk of a type II enor, I used a p < 0.1 to allow nteracton terms to reman n the model. However, for the fnal model the tolerance for type I error was set at 0.05 for man effects. 1 tested whether overall breedng success of certan nest stes was nfluenced by the same nest-ste characterstcs that prevented gull attacks or not. For ths analyss my sample sze conssta of the same 300 randomly chosen nest stes than n the man analyss (see above). 1 used a generazed lnear model wth a bwy response varable (successful or not successfd n rasng a chck), a logt lnk functon, and sx ndependent varables: plot, ledge wdth, roof, number of walls, densty and CM part. 1 folowed the same method for fndng the best f* model than descrbed above. Melane Mzsssmo - Lmgc gull pndaton m &lad-legged kttzookes 67

87 For the next set of analyses 1 chose randomly 300 nests out of 617 actve nests from three study plots (N4,S and P2). 1 had four response varables: attacked or not attacked (1) by herrng gus under calm condtons, (2) by great black-backed gulls under calm condtons, (3) by herrng gulls under wndy condtons and (4) by pat black-backed guh under wndy condtons. Plot S1 was excluded from ths analyss because only herrng gulls foraged at ths plot under calm condtons. 1 dd four analyses, one for each response varable, wth followng sx ndependent varables: plot, ledge wdth, roof, number of wak, densty and clff part. I used generalzed lnear models and logt lnk functons and followed the same procedure as desabed above for fndng the best fttng model. In cases where there were no attackç on nest stes wth a certan characterstc, those terms were not estmable and dropped out of the fnal model. The term 'sgnfcance' s used n relaton to statstcd tests and does not mply bologcal mportance.

88 3.4. Results Dfferences n gul predaton rsk and kttwake breedng success among plots ncludng au nest stes For each plot the number of actve nests, the percentage of attacked nests and successful nests s reported n Table 3.3. The proporton of nests attacked dffered sgnhcantly among plots ( ~ = , df = 3, p < ). Plot SI had the hghest percentage of attacked nest stes, followed by FZ,N4 and S5 (Table 3.3). 1 receved the same results n the analyss ndudng 300 randomly chosen stes (see below). The percentage of successful nests n 1998 or 1999 vared sgnfcantly among plots ( ~ = , df = 3, p c ). Plot SI whch had the hghest proporton of attacked nests (see above), also had the lowest percentage of stes that rased dcks (33.3%, 14/42), folowed by plot P2 wth 55.6% (45/81), and plots S5 and N4 had the hghest percentage of 72.4% (194/268) and 76.6% ( ) respectvely (Table 3.3). 1 also receved the same results n the analyss ncudng 300 randomly chosen stes (see beow). Melune Masssaro - Large gull pndntm on blrtdr-legged ktttoaks 69

89 Table 3.3. Nest numbers, percent nets attacked by h d g or great blackbacked gulls and percent nets successful n rasng at least one ch& n four kttwake plots on G d Island n 1998 and Kttwake Number of nests that O/O nests O/O nests successfu study plot were actve (eggs attlcked (attacked (successful nesw lad) n 19!38 or 1999 nesq actve nests) actve ne&) Melane Ahssmo - Lge guu prrdutm on bu-egged kttnuakrs 70

90 3.4.2 Gd attacks and nest-ste characterstcs In the man analyss ncludng 300 randomly chosen nest stes, plot, densty and clff part had a sgrufcant effect on whch nest stes were attacked by gulls (Table 3.4). Nest stes n densty areas were more lkely to be attacked (33.0%; 38/115) than stes n hgh (30.4%; 14/46) or low (15.8%; 22/139) densty areas. Stes at upper parts of dffs had a hgher lücelhood of kng attacked by large gulls (45.2%; 38/84) compared to nest stes located at mddle (20.5%; 23/112) or lower parts of the clff (12.5%; 13/104) Kttwake breedng success and nestate characterstcs Plot, ledge wdth and densty affected sgnhcantly where chcks fledged (Table 3.5). Of 300 randomly chosen nests, stes on narrow ledges had a hgher lkelhood of succeedng n rasng chcks (73.0%; 135/185) than stes on broad Iedges (59.1%; 68/115). Stes located n hgh densty areas had a hgher success rate n rasng chcks (ï3.9%; 34/46) than stes n (68.7%; 79/115) or low (46.3%; 90/139) densty areas.

91 Table 3.4. Generalzed lnear model of kttwake nest-ste charac terstcs that reduced the rsk of predaton attacks by herrng or great black-backed gulls on Gd Island n 1998 and 1999 (n = 300 randomly chosen stes). Source df F P PLOT 3 LEDGE WIDTH 1 ROOF 1 NO. WALE 1 DENçITY 2 CLFF PART 2 Note: Only actve kttwake nesb (2 1 egg was lad) were ncluded n the analyss. Hgher order terms not present were > 0.1) and dropped from the model. For the fnal model the tolerance for type I error was set at 0.05 for man effecf~. Generalzed Inear model wth a bnary response varable and a logt lnk hcton. Melane Masssmo - Large pl1 prr&tîun m bhk-legged htwcrkes 72

92 Table 3.5. Generalzed lnear model of kttwake nest-ste characterstcs that nfluenced the breedng success of kttwakes on G d Island n 199û and 1999 (n = 300 randomly chosen stes). Source df F P P m 3 LEDGE WIDTH 1 ROOF 1 NO. WALLS 1 DENSITY 2 CUFF PART 2 Note: Only actve kttwake nests (>_ 1 egg was lad) were ncluded n the analyss. Hgher order terms not present were nsgnfcant (p > 0.1) and dropped from the model. For the fnal model the tolerance for type 1 enor was set at 0.05 for man effects. Generazed near model wth a bnary response varable and a logt lnk functon. Melune Mmssaro - Large gull predaton on back-legged kttroakes 73

93 Cornparson of foragng decsons of h d g gulls and great black- backed gulls n relaton to whd condtons Durng calm condtons, roofs over nest stes, the locaton on the clff as well as ledge wdth reduced the rsk of herrng gull attacks on kttwakes (Table 3.6). Of al nest stes wthout roofs, 14.0% (18/129) were attacked, however only 6.4% (î/ln) of al1 nest stes wth roof were attacked. Nest stes on upper parts of clffs had a hgher lkelhood of kng attacked by herrng gulls (15.9%; 13/82) than stes on mddle (11.0%; 12/109) or lower parts (3.7%; 4/109). There was a sgrufcant nteracton effect among plots and ledge wdth. Whereas at plot S5, stes located on narrow ledges experenced a hgher rsk of predaton by herrng gulls durng calm condtons (16.9%; 21/65; compared to stes on broad ledges: 7.4%; 4/54), stes on broad ledges had a hgher rsk of predaton at plot M and N4 [M: stes on broad ledges: 20.0% (2/10); stes on narrow ledges: 6.3% (2/32); N4: stes on broad ledges: 13.7% (7/51); stes on narrow ledges: 3.4% (318m Durng calm condtons, plot, ledge wdth and clff part affectec sgrufcantly whch nest stes were attacked by great bladc-backed gulls (Table 3.6). Nneteen percent of nest stes at plot P2 were attacked by great black- backed gulls (8/42), however only 7.9% (11/139) at plot N4 and 3.4% (4/119) at plot S5. Stes on broad ledges experenced a hgher rsk of predaton by great Melune Masssmo - Lmgc pl1 predntan on bladr-legged cttroakes 74

94 black-backed gulls (13.0%; 15/115) than stes on wrow ledges (4.3%; 81185). Nest stes on upper parts of clffs had a hgher lkelhood of kng attacked by great black-backed gulls (17.1%; 14/82) than stes on mddle (4.6%; 5/109) or lower parts (3.7%; 4/109). Durng wndy condtons, the locaton on the clff affected the rsk of predaton by both hemng and great black-backed gulls (Table 3.6). Stes located on upper parts of clffs had a hgher lkelhood of beng attacked by herrng and great black-backed gulls 12.2 % (10182) and 13.4% (11/82), respectvely] than stes on mddle [0.9% (1/109) and 1.9% (2/109), respectvely] and 1owe.r parts [1.9 % (21 109) and 3.7% (41 log), respectvely 1. However, densty also affected the rsk of predaton by great black-backed gulls n wndy condtons (Table 3.6). Stes n densty areas had a hgher lkelhood of beng attacked by great black-backed guls (10.5%; 12/114) than stes n low (2.1%; 3/140) or hgh densty areas (2.2%; 1/46).

95 Table 3.6. Four generalzed lnear models, testng whether certan kttwake nest-ste characterstcs nfuenced whch stes were attacked by (1) herrng gus (= HERG) durng calm condtons, (2) great black-backed gulls (= GBBG) durng calm condtons, (3) h k g gulls durng wndy condtons, and (4) great black-backed gulls durng wndy condtons. Nest-ste characterstcs Cam condtons Wmdy condtons HERG GBBG HERG GBBG PLOT IIS O LEDGE MrIDTH ns ns ns ROOF s 11s ns NO. WALLÇ ns - IIS IIS DENSITY lu m CLEF PART PLOT * LEDGE MrIDTH O O - Note: In cases where there were no attacks on nest stes wth a certan characterstc, those terms were not estmable and dropped out of the fnal model. Only sgntfcant p values are reported. Melane Masssmo - Lmge gull p&tm a blneklegged kttwakes 76

96 3.5. Dscusson The effect of plot sze and nestng densty on nest predaton and kttwake breedng success I found sgnfcant dfferences n the proportons of ktawake nests attacked by gulls and kttwake breedng success among plots. On Gd Island, ndvdual kttwake nests at the smallest plot SI experenced a hgher probablty of kng attacked by herrng or great black-backed gulis than at larger plots. Hence, the percentage of faled nests was hghest at plot S1 and decreased as the sze of the plots (number of nests) ncreased. The number and foragng ablty of resdent breedng guus, whch occuped kttwake nestng cffs as feedng terrtores, mght explan the varablty of nest predaton among plots. Although 1 obsewed study plots for many hours, there was some evdence that 1 mght have montored the foragng behavour of only a few large gulls. In 1999 on Gu11 Island, we unquely colour banded one male great black-backed gull, whch held a feedng terrtory at kttwake study plot N4. Sxty percent of al1 great black-backed gu predaton attempts on kttwakes at N4 were made by ths ndvdual (M. MassaroI unpubl. data). In brd speces that show mobbng behavour as a defense strategy aganst predators, nestng n large, dense colones offers advantages aganst

97 predators (Wttenberger and Hunt 1985). Several studes have show that f the rsk of predaton s hgh, cw-nestng black-legged kttwakes defend ther nests by vgorous mobbng aganst avan predators (Andersson 1976; Montevecch 1979; Maccarone 1992). Durng ths study 1 frequently observed that kttwake eggs and chcks were not only defended by ther own parents, but aso by cooperatve mobbng of prospectors as well as faled nesters. Several tmes 1 observed that kttwakes made physcal contact wth an attackng great black-backed gul n order to hnder the gull from landng or remanng on the ledge. At a declnng thck-blled murre colony glaucous gulls were more lkely to forage on foot on broad ledges, where, because of populaton declnes, murres nested at lower denstes (Glchrst 1999). However, at hghly populated and dense murre colones glaucous gulls were gven less opportunty to forage on foot than n low densty murre ledges. Surprsngly n ths study, nest stes wth two actve neghbours ( densty) were more kely to be attacked than stes wth less or more actve neghbours. Breedng success (percentage of successful nests) was hghest n hgh densty areas and lowest n low densty areas. Ths suggests that t s most advantageous for a kttwake par to breed n a hgh nestng densty area wthn a large sub-colony where predaton pressure per ndvdual subcolony member was lower than n small sub-colones, ncreasng the chance of Melane Masssaro - Lmge gull predutcm on blrtck-legged kttwkes 78

98 reproductve success. However, two questons reman unanswered: (1) Why were gulls more lkely to attack nests n densty areas than n low densty areas? and (2) Why was breedng success substantally lower n low densty areas than n densty areas, although predaton pressure s paradoxcally hgher n densty areas? Foragng decsons of gulls mght be nfhenced by the trade-off between maxmzng energy gan and mnrnzng rsk of njury (Sten 1977; Glchrst et a. 1998). Gulls may have found the optmum foragng tactc by attackng stes n densty areas, where the level of mobbng behavour s tolerable and the possble energy gan, n case the gull succeeded n landng on the ledge, substantally hgher than n low densty areas. The low breedng success n low densty areas could be attrbutable to the lower qualty of brds breedng at the edge of a subcolony (Coulson 1968) or to a lower level of socal stmulaton that may cause low breedng success (Danchn 1988). Vewed n a broader perspectve, a hgh percentage of faled nestng attempts experenced by a small sub-colony cm have long-term effects on recmtment. Frst-trne breedng kttwakes as wd as adults choose ther nestng locaton basec partly on ther own reproductve success and that of conspecfc nesters durng the prevous breedng season (Danchn and Monnat 1992; Cadou et al. 1994; Danchn et al. 1998). Low average reproductve success

99 at smd sub-colones, such as SI, mght provde an ndcaton to pre-breedng kttwakes on the local qualty of the sub-colony, causng rdtng kttwakes to choose larger clffs (Danchn et al. 1998). Furder confrmaton of ths possblty at Gull Island would requre replcated small versus large plot cornparsons Relatonshps behveen kttwake nest-ete characterstcs, gull predaton and kttwake breedng success Besdes a sgrufcant plot and densty effect, 1 found that nest stes at the upper parts of clffs experenced a hgher probabüty of kng atgcked than stes located at the mdde or lower parts of clffs. Smlarly, Maccarone (1992) observed that ravens on Baccaleu Island, Newfoundland, hunted dong the upper thrd of a kttwake nestng clff on 49% of al patrols, 33% along the mddle and 18% along the lower thrd of the clff. At a thck-blled murre colony, Gaston and Ellot (1996) found that 68% of al predaton attempts by ravens occurred n the upper 30% of the clff, although al nestng stes were almost evenly dstrbuted n relaton to the dstance from the upper edge of the clff. They also observed that wthn the top 30% of the dff, perîpheral stes were more lkely to be attacked than central stes. In an ear1er study on Great Island, Wtless Bay, kttwake nest-ste poston relatve to the perphery of the

100 colony dd not dffa between nests wth clcks and random nests (Regehr et al. 1998). Instead they found that nets wth chcks had larger overhan~ than random nests. In ths study, an overhang (roof) over a nest ste reduced only the rsk of predaton by herrng guls durllg calm condtons. When 1 tested whether breedng success of certan nest stes was Ilfluenced by the same nest-ste characterstcs that reduced the rsk of gull attacks, 1 found that plot and densty also affected breedng success. Whereas, gull attacks were amed more lkely at stes located on upper parts of clffs, 1 dd not fnd any breedng success dffaences among upper, mddle and lower parts. However, ledge wdth nfluenced the breedng success rate of nests, whereby stes on narrow ledges showed a hgher percentage of success. These results suggest that although reproductve performance of a par of kttwakes was lkely nfuenced by nest-ste characterstcs reducng the rsk of predaton, also a wder varet- of factors, ncludng qualty of the par, age, and paraste abundance may nfluence breedng success Relatonshps of wnd conâtons and foragng decsons of guls If foragng decsons are nfluencec by the hade-off between possble energy gan and the rsk of njury (Clchrst et al. 1998), n partdar, opportunstc foragers, such as herrng and great black-backed gulls shoud be confronted Melane Mnsssaro - Lmge gull predatm on black-kgged ktîwpkes 81

101 frequently by those decsons, as they are capable of swtchng to dfferent prey (Perott and Annett 1987). Condtons that constran the foragng abüty of predators are hghly dynamc due to changes n prey avalablty, competton among predator speces, and envronmental condtons (Verbeek 1977; Bard 1990; Van Heezk 1990; Anderson and Hodum 1993; Glchrst et al. 1998). Several studes have shown that avan predators respond to dfferental wnd condtons by changng ther foragng W c s (Spear and Anderson 1989; Young 1994; Glchrst and Gaston 1997). Although both h-g gulis and great blackbacked gulls have a wng loadng of 48 N/mz (Pennycuck 1987; Spear and Anley 1993, ther foragng tactcs dffer due to ther sze. Regehr (1994) obsewed that kttwakes usudy left ther nests when a great black-backed gull soared above them and 15% of eggs taken were depredated by huntng ths way. However herrng gulls never took any eggs ths way (Regehr 1994). On Gull Island, I obsewed that breedng kttwakes usualy stayed on theù nests even when a great black-backed gull soared above. Regardless of wnd condtons, herrng and great black-backed gulls attacked to a greater percentage nest stes located at the upper parts of clffs than at mddle and lower parts. However, herrng gulls had more dffdty foragng on nest stes wth roofs durng c h condtons. Herrng gulls were observed to start most of ther foragng attacks on kttwakes from the upper edge of the dff. From Melune hhsssaro - Large guil pndaton on black-egged kttrp&s 82

102 that poston they ether tned to wak nto the kttwake colony or they jumped nto the ar flyhg n a srnall 180 sem-crcle before attackng a ste. Usualy herrng gulls stole kttwake eggs or chcks by supportng ther own weght by rapd wng beatng and lowerng ther feet on the kttwake ledge. Sometmes hehg gds removed adult kttwakes from the nest before nest contents were taken. The foragng effort of great black-backed gulls was constraned by narrow ledges durng calm wnd condtons. In contrast to herxng gulls, great black-backed gdls started most of ther attacks by flyng crcles dong the kttwake nestng clff. Usually they lowered ther flght speed and then tred to land on one of the kttwake ledges. Once they landed successhlly on a ledge, great black-backed gulls waiked wthn the kttwake colony robbng a nests they could reach on the ledge. Wndy condtons lkely hcreased the aeral maneuverablty of both gull speces. Great black-backed guîls were able to land more successwy on kttwake ledges and then attadc nets on foot. They chose to attack a sgnfcant hgher proporton of stes wthn dewty areas than stes n low or hgh densty areas. However, herrng guîs, kng more mnerable to kttwake defense behavow due to ther smaîler sze, were observed to forage on the wng, almost standng s a n md-ar over a kttwake clff, and perfom sudden attempts to steal an egg or ch& from a nest wthout landng on a Melune Masssaro - L qe gull pdatm un blpck-lggcd ktîwakes 83

103 ledge. Montorng ndvdual gull foragng behavour and measurng wnd condtons ncludng up and down drafts at dfferent devatons of a dff mght gve more dues as to why large guls prefer to forage on certan nest stes. In concluson, among my study plots at Gul Island, nests at smaller kttwake sukolones experenced a hgher rsk of kng depredated by gulls than nests at larger ciffs, resultng n a lower breedng success at s d dffs. Gu11 attacks were more frequent on stes n densty areas than on nest stes n low or hgh densty areas. Nests located at the upper part of the CW experenced a hgher probablty of kng attacked than nests at the mddle or Iower part of the clff. Breedng success was correlated wth ledge wdth and densty and vared sgmfcantly among plots. Wnd condtons nfluenced whch nest-ste characterstcs reduced the rsk of predaton. Further nvestgaton on how physcal clff structures constran the foragng ablty of avan predators should focus on three man ssues: (1) large sale clff characterstcs, ncludng a larger sample sze of nestng clffs, (2) wnd condtons at dfferent elevatons of the clff, and (3) predator dynamcs, such as breedng densty, competton, and foragrtg range. Melane Mzsssm - hge gull pndatm on black-kgged kftwukes 84

104 3.6. References Andersson M (1976) Populaton ecology of the Long-taled Skua (S ter~~ranus longcnudus Veül). J Anh Eco1 45: Anderson DJ, Hodurn PJ (1993) Redator behavor favors clumped nestng n an oceanc seabrd. kology 74: Bard PH (1990) nfluence of abotc factors and prey dstrbuton on det and reproductve success of three seabrd speces n Alaska. Orns Scand 21: Barbraud C (1999) Subspeces-selecfve predaton of snow petrels by skuas. Okos 86: Barrett RT, Runde OJ (1980) Growth and survval of nestlng Kttwakes Rsw hdactyla n Norway. Orns Scand 11: 22û-235 Brkhead TR, Greene E, Bggns JD, Netteshp DN (19û5) Breedng ste characterstcs and breedng success n Thdc-bed Mmes. Can J Zoo1 63:188O-I884 Burger J, Gochfeld M (1994) Predaton and effects of hurnans on sland-nestng seabrds. BrdLfe Conservaton Seres 1: 3947

105 Cadou B, Monnat JY, Danchn E (1994) Prospecthg n the kttwake, Rssu bdactyln: dfferent behavoural patterns and the role of squatkg n recmtment. Anm Behav 47: Cavanagh PM, Grffn CR (1993) Responses of nestng cornmon tems and laughng gulls to flyovers by large guils. Wlson Bull 105: Coulson JC (1963) The status of the Kttwake n the Brtsh Isles. Brd Study 10: Coulson JC (1968) Dfferences of the qualty of brds nestng n the centre and on the edges of a colony. Nature 2W Cullen E (1957) Adaptatons n the Kttwake to dff nestng. Ibs 99: Danchn É (1988) Socal nteractons n kttwake colones: socal facltatow and/or favouable socal envronment. Anm. Behav. 36: Danchn É, Monnat J-Y (1992) Populaton dynamcs modelng of two neghbourng kttwake Rssa tndacfyln colones. Ardea 80: Danchn É, Bouüner T, Massot M (1998) Conspecrhc reproductve success and breedng habtat selecton: mplcatons for the shdy of colonalty. Ec010gy 79: Gaston AJ, Nettleshp DN (1981) The Thck-blled Murres of Prnce Leopold Island. Can Wldl Serv Monogr No 6 Melane Masssaro - toge gull pndatm on black-legged kttwalus 86

106 Gaston AJ, EIlot RD (19%) Predaton by Ravens Coms anax on Brunnch's Gullemot UM lomûu eggs and ch& and ts possble mpact on breedng ste selecton. Ibs 138: Glchrst HG, Gaston AJ (1997) Effects of mme nest ste characterstcs and wnd condtons on predaton by glaucous gulls. Can J Zoo1 75: Glchrst HG, Gaston AJ, Smth JNM (lm) Wnd and prey nest stes as foragng constrants on an avan predator, the glaucous gull. Ecology 79: Glchnst HG (1999) Declnng thck-blled murre UM bmoa colones experence hgher gul predaton rates: an nter-colony cornparson. B01 Cons 87: Lack D (1954) The natural regulaton of anmal numbers. Clarendon Press, Oxford, UK Lock AR, Brown RGB, Gerrets SH (1994) Gazetteer of marne brds n Atanaf Canada. Canadan Wldlfe Servce, Atlantc Regon, 137pp. Maccarone AD (1992) Predaton by Common Ravens on Clff-nestng Black- legged Kttwakes on Baccaleu Island, Newfoundland. Col Waterbrds 15: Martn TE (1988) Habtat and area effects on forest brd assemblages: s nest predaton an nfluence? Ecology 69:74-84

107 Martn TE (1993) Nest Predaton and Nest Stes. BoGcence 43: Massaro M, Chardne JW, Jones IL,, GJ Robertson (2000) Delayed capeln (Mallotus vliosus) avaüablty nfluences large gull predatory behavour on black-legged kttwakes (Rssa tn'dactyla), causng a reducton n kttwake breedng success. Can J In press. Maunder JE, Threlfd W (1972) The breedng bology of the black-legged kttwake n Newfoundland. Auk 89: Montevecch WA (1979) Predator-pre y relatonshps between ravens and kttwakes. Zetschrf t Hlr Terpsychologe 49: Murphy EC, Sprnger AM, Roseneau DG (1991) Hgh annual varablty n reproductve success of kttwakes (Rssa hr*dacfyla L.) at a colony n western Alaska. J Anm Eco1 60: Pe~ycuck CJ (1987) Flght of Auks (Akdae) and other northem seabrds compared to southem Proce~arformes: Omthodolte observatons. J Exp Bol12û: Perott R, Annett C (1987) Reproductve consequences of detary specalzaton and swtchng n an ecologca generalst. In: Kaml AC, Krebs JR, Pullam HR (eds) Foragng behavour. Plenum Press, New York, p Mefane Mnssmo - Large gull pdatm on blodr-kgged kttwokes 88

108 Rachlow JL, Bowyer RT (1998) Habtat selecton by Dal's sheep (Oos dnll): matemal trade-offs. J Zoo1 Lond 245: Regehr HM, Montevecch WA (1997) Interactve effects of food shortage and predaton on breedng falure of black-legged kttwakes: ndrect effects of fsheres actvtes and mplcatons for ndcator speces. Mar Eco1 Prog Ser 155: Regehr HM, Rodway MS, Montevecch WA (1998) Antpredator benefts of nestste selecton n Back-legged Kttwakes. Can J Zoo1 76: Shealer DA, Burger J (1992) Dfferenüal responses of tropcal roseate terns to aeral nûuders throughout the nestng cycle. Condor 94: 7îZ07''l9 Spear LB, Anderson DW (1989) Nest-ste selecton by Yellow-footed Guls. Condor 91: Spear LB, Anley DG (1997) Flght behavour of seabrds n relaton to wnd drecton and wng morphology. Ibs 139: Sten RA (1977) Selectve predaton, optmal foragng, and the predator-prey nteracton between fsh and cra yfsh. Ecology 58: Tuck LM (1961) The mumes. Cm Wdl Serv Monogr No 1 Van Heezk Y (1990) Dets of ydow-eyed, Fordand crested, and lffle blue penguns breedng sympatrcally on Codfsh Island, New Zealand. New Zealand J Zoo1 17:

109 Verbeek NAM (1977) Interactons between h b g and lesser black-backed gulls feedng on refuse. Auk 94: Wttenberger JF, Hunt GL (1985) The adaptve sgnfcance of colonalty n brds. In Avan Bology Vol VIII, edted by Famer DS, Kng JR, Parkes KC, Academc Press, Orlando. pp 1-78 Yoro Pt Quntana F (1997) Predaton by Kelp Guls Lmcs domnmus at a mxed-speces colony of Royal Tems Sterna mmma and Cayenne Terns Stema euryptha n Patagona. Ibs 139: Young E (1994) Skua and Pengun: predator and prey. Cambrdge Unversty Press, Cambrdge. Melane Massswo - h ge gu1 p&ton un biack-zegged kttwakes 90

110 Fnal Dscusson and Conclusons The man objectve of ths study was to nvestgak the causes and effects of herrng and great black-backed gul predaton on kttwakes. I approached the topc of ths shdy from two dfferent angles: (1) from a coarse sale perspectve ncludng nter-trophc relatonshps and the mplcatons of large gull predaton on kttwake populatons n Wtless Bay and (2) from a fne sale perspectve examnng kttwake plot dffaences, nestng densty, nest-ste characterstcs, local wnd condtons as wd as the rsk of predaton for any ndvdual kttwake nest. In ecosystems where one speces, such as capeln, s a predomnant prey tem for seabrds, nter-trophc relatonshps are relatvely clear and hence offer an opportunty to study the effects of reduced avalablty of a marne prey speces on sea brd populatons. The results of ths study strongly suggest that the tmng of nshore spawnng of capeln nnuenced the predatory behavour of large gulls on kttwakes and hence kttwake breedng performance. Kttwakes have been affected by the delayed arrva1 of capeln n recent years bot. ndrectly (due ncreased predaton by gulls) and drecty (due to reduced food avalablty; Regehr 1994). Commercal fshng or abotc factors, such as Melane Masssaro - LRge gull predaton on bacù-legged kffzoakes 91

111 change n water or ar temperatures, often cause reduced marne prey avalablty and decrease seabrd breedng performance (e.g. Sprnger et al. 1984; Hatch 1987; Anderson 1989; Bard 1990; Hamer et al and 1993; Murphy et al. 1991). For example, kttwake populatons n Alaska showed hgher reproductve success n breedng seasons followng wann sprngs and reduced success followng cold sprngs (Murphy et al. 1991). The sprng of 1999 was one of the warmest of ths decade n Newfoundland. In 1999, when capeln arrved 9 days earler to spawn, kttwake breedng success on Gd Island was 53.3% hgher than n Warm sprng temperabes may gve breedng kttwakes an ndcaton of the food avalablty durng ths season and may affect egg-layng and clutch sze. In my study, large gull predaton was hghest when gulls had small chcks to feed but capeln was not avalable. In 1998, an nland kttwake plot (not ncluded n ths study) on Gd Island was about a week delayed n breedng n comparson to other kttwake plots (J. W. Chardne, unpubl. data). It seemed that ths plot benefted from ths delay, because only a smau proporton of kttwake eggs were lad n the perod of hgh gull predaton. In years of late capeln avalablty, t could be advantageous for kttwakes to delay breedng. However, n the study of Coulson and Thomas (1984) breedng success declned f eggs were lad after the frst thrd of the breedng season. In Mehe Masssmo - Larp gull pàatm un black-legged kttropkes 92

112 1969 and 1970 mean egg-layng dates for kttwakes on Gd Island were 3 June and 29 May, respectvely (Maunder and ThreUal1972). Mean egg layng occurred around the same tme n 1998 and 1999, suggestng that kttwakes dd not delay breedng due to latex capeln avalablty. It s questonable how quskly and effcently seabrds cm adapt ther lfe hstory strateges to large scale envronmental changes. In chapter 2 and chapter 3 1 looked at dfferences among plots on Gu11 Island: mean gull predaton attempt rates were lowest at plot SI, the smallest kttwake nestng aggregaton, however the hghest percentage of nests were attacked at plot SI. Predatoa are attracted to seabrd colones because a large concentraton of food s avalable (Wttenberger and Hunt 1985). Whereas, on Gu11 Island, the smalest kttwake plot SI was occuped by only one herrng gull par, the Iargest plot SS was occuped by one great black-backed gull par and one hemng gull par. It appears that larger kttwake sub-colones attracted more and large predaton (great versus whch requre a larger feedng temtory to be able to adequately feed ther Young. However, clearly the predaton pressure for an ndvdua kttwake nest decreased as the sue of the subolony ncreased. The percentage of nests where chcks fledged was more than twce as hgh at large plots (N4 and S5, compared to plot SI). The resdts of ths study suggest that predators were Melane Masssmo - Large gull prrdatm on biack-kgged httwakes 93

113 attracted to larger sub-colones than srnalfer ones, however the predaton rsk per ndvdual colony member s decreased n larger sub-colones. The effect of ncreased populatons of large gulls on other seabrds has been controversal for several decades. There s a wdespread opnon among the general publc n Newfoundland Uat gulls are pest speces due to theh huge presence near landfll stes and around fshng operatons. Whereas several studes have shown that large gulls have been responsble for declnng seabrd populatons (e-g. Hatch 1970; Gldrst 1999; Whttam and Leonard 1999), other studes pont out that even when pl predaton or kleptoparastsm s evdent t has lttle negatve effect on prey populatons (e.g. Perott 1983; Rce 1985; Cavanagh and Cran 1993; Howes and Montevecch 1993). For Gull Island, I estmated that gulls took 43% and 30% of al1 kttwake eggs lad n 1998 and 1999 respectvely. However, t s unknown wnether gull predaton has an overd negatve effect on kttwake populatons n Wtless Bay. As ponted out earler, t appeared that only a few resdent breedng gu1 pars were responsble for most kttwake predaton. When predatory gulls were removed n a study of golden plovers (Pluoals (tf117'arn). plover numbers dd not ncrease (Parr 1993). 1 predct a smlar eftect on GuU Island f resdent breedng gds were removed. After removng resdent gull pars, predaton rates rnght decrease for a few weeks, but 1 predct other gulls, n partdm

114 recrutng gulls, would soon occupy the feedng temtory and prey upon kttwakes. If guu predaton becomes an evdent problem for seabrd populatons n Newfoundland, 1 suggest approachng the problem at the source and tryng to alter human behavour nstead of 'blamng' gulls. In Newfoundland ths may requre a change n managng fsh and household waste and reducng the quota for the annual capeln fshery Ref erences Anderson DJ (1989) Dfferental responses of boobes and other seabrds n the GalApagos to the El No- Southem Osdaton event. Mar Eco1 Prog Ser 52: Bard PH (1990) Influence of abotc factors and prey dstrbuton on det and reproductve success of three seabrd speces n Alaska. Oms Scand 21: Cavanagh PM, Grffn CR (1993) Responses of nestng common tems and Iaughng gulls to flyovers by large gulls. Wlson Bull 105: Coulson JC, Thomas C (1984) Dfferences n the breedng performance of ndvdual kttwake gulls, Rssa tràuctya (L.). In Behavourd Ecology - Melane Masssmo - Lmge gull prcdoîm on black-egged kttwokes 95

115 EEologcal Consequences of Adaptve Behavour, edted by SMy RM, Smth RH, Blackwell Scentfc Publcatons. pp Glchrst HG (1999) Declnng thck-büled murre UM IamPa colones scperence hgher gull predaton rates: an nter-colony cornparsun. B01 Cons 87: Hamer KC, Furness RW, Caldow RWG (1991) The effects of changes n food avalablty on the breedng ecology of great skuas Cathmacta skua n Shetland. J Zoo1 Vnd) 223: Hamer KC, Monaghan P, Uttley JD, Walton P, Burns MD (1993) The nfluence of food supply on the breedng ecology of Kttwakes Rssa h'dactyla n Shetland. Ibs 122: Hatch JJ (1970) Predaton and pkacy by gulls at a temery n Mane. Auk 87: Hatch SA (1987) Dd the El Nfe Southem Oscllaton affect seabùds n Alaska? Wlson Bull 99: Howes LA, Montevecch W A (1993) Populaton trends and nteractons arnong terns and gulls n Gros Morne Natonal Park, Newfoundland. Can J Zoo1 n: Maunder JE, Threlfall W (1972) The breedng bology of the black-legged kttwake n Newfoundland. Auk 89: Melane Masssmo - Lmge gufl pdpttm on block-kg@ kltzaakes %

116 Murphy EC, Sprnger AM, Roseneau DG (1991) Hgh annual varablty n reproductve success of kttwakes (Rssa fnfndactyla L.) at a colony n western Alaska. J Anm Eco1 60: Parr R (1993) Nest predaton and numbers of Golden Plovers Pluvals ~'carn and other moorland waders. Brd Study 40: Perott R (1983) Gull-puff nteractons on Great Island, Newfoundland. B01 Cons. 26: 1-14 Regehr HM (1994) Breedng performance of black-legged kttwakes on Great Island. Newfoundland, durng perods of reduced food avalablty. MSc thess, Mernoral Unversty of Newfoundland, St. John's Rce J (1985) Interactons of varaton n food supply and kleptoparastsm levels on the reproductve success of Cornmon Puffns (Frutermln arctca). Can J Zoo1 63: Sprnger AMI Roseneau DG, Murphy EC, Sprnger MI (1984) Envronmenta controls of marne food webs: food habts of seabrds n the eastern Chukch Sea. Can J Fsh Aquat Sc 41: Whttam RM, Leowd ML (1999) Predaton and breedng success n roseate tems ( Stm hgall). Can J Zoo1 77: Melane Masssaro - Large gu2 pndatm on blpck-legged kttrvaùes 97

117 Wttenberger JF, Hunt GL (19û5) The adaptve swcance of colonalïty n brds. In Avan Bology Vol WI, edted by Famer DS, Kng JR, Parkes KC, Academc Press, Orlando. pp 1-78

118

119 Cul1 Island 47' 16' N, 52" 46' W O 300 kzkkd Study plots A Weather staton Newfoundland 1 Locato n of and

120 Appendx 2. Kttwake nest-ste characterstcs for each nest ste at four study plots (SI, P2, N4 and S.) on Gd Island n 1998 and 1999; P = Plot, NS = Nest Ste, LW = Ledge Wdth, R = Roof, W = No. of Walls, D = Densty, CI? = Clff Part, and A = Actve. Last column ndcates whether at least one egg was lad n ths nest ste n 1998 or 1999 (actve nest ste = 1) or not (= O). Please see for Mer detaled defntons of nes t-ste charaderstcs secton Nest-ste characterstcs on p. 60. P NS LW R W D CP A ST 1 broad None 2+ d u m O S 2 braad Roof S1 3 b d W S14narrowRoof SI S broad None SI 6 broad None S 7 broad Roof S sl 8 narrow Roof 9 nanow Roof SI 10 broad Roof SI 11btoed Ra3f S 12 btoed None S 13 m w S S S SI SI Roof 14 narrow None 15 narrow Rmf 16 narrow None 17 namw None 18 narmw None SI 19 bmd None S1 2û bmd None S1 2î narrow None SI 22 broed None S1 sl S 23 narrow Rmf 24 narrow None 25 nam;>w None S 26 bnad Roof S 27 broad None low 1-2+ low low 1-1- hgh 2+ Low hgh mdum 1- bw duut 2+ low 1- h@ upp= lower UPF medjum upp= upp= U P F P NS LW R W D CP A Sl 28 ~ n o w None 2+ ' 1- low law low 1- low 1-1- low 1- h& 2+ bw 1- bw 1- low 1- bw 2+ low 2+ &w 2+ low 1-1- h& 1- bw bw 1- bw 1- bw Melane Masssmo - Large gull pdnton on black-egged kttwakes 101

121 P2 1 narrow F2 2 namw P2 3 m w Fz 4 narrow P2 5 b d PZ 6 baad P2 7 mow P2 8 broad P2 9 h d P2 10 nanow Rmf P2 11 narrow Roof P2 12 narrow Roof P2 13 MKOW None P2 14 narrow None P2 15 narrow None P2 16 h d fz 17 nmow P2 18 nanow P2 19 narrow PZ 20 Mnow ~2 n -W P2 22 h d P2 23 bmad PZ 24 brcnd PZ 25 bmad P2 26 nanow P2 27 nanow P2 28 narrow PZ 29 narrow PZ 30 broad P2 31 brmd P2 32 h d P2 33 m w P2 34 narrow PZ 35 m w narrow None 1- upper O (PZ 36 nurow Rd 2+ Lowa 1 Sl 6 narrow None 1- low nurpw Rd 2+ S 66 bmad None 1- medjum numw Root 1- S 69 braad None 1- bw nurow kd 1- bw R d 1- bw 1- hgh Roof 1- meùum 1- hgh R4x)f None None None Roof RDof None None Roof None None Roof Raf Rmf Roof Roof Roof None Rn04 R d Rad Roof Roof Roof Roof Roof Roof 1- low 1- bw 1- km 1- low 1-1- hgh 1- hgh 1- hgh 1- hgh 1-1- Low 1-1- bw 1- low bw 1- dm h& 1- h& 1-1- kw 1- kw 2+ bw 1- kw 1-1- low 1- hgh 2+ bw medjum lowe nurow Non bclnd None bod Ibd bnsd Roat blod Roa kod R d bmed Roa bpd Roa 1\IvroW ROOt numw None nurow Rd nurow Rod nurow R d nurow R d numw R d nmw Roa m w Rad narrow R d nurow Rod numw None nurow ROOt nurow None nurow Roa kmd Nom blod None nurow Nom nurow Nme nuraw None nurow Non nrrrow Non nutow None nurow Nom Mrrow Root MmnwRoot mxmw w 1- hgh bw medjum 1- bw 1- low 1- bw 1-2- law 1- low bw bw 2+ bw 1- bw 1- law 1- bw 1-1- d u m 1- Lnedum 1- zndum bw 1-1-

122 PZ 76 namw Roof ~2 n n mw Roor PZ n m w m P2 79 nanow R d P2 80 h d Roof PZ 81 ûroad Raof P2 82 m w Roof Pî 83 b d Rad F'2 84 narrow Rœtf P2 85 narrow None P2 86 narrow F&uf P2 87broed R d PZ 08 bmd w P2 90 narrow None P2 1 nanow None P2 92 narrow Roof P2 93 nmow Roaf P2 94 narrow None P2 95 broad P2 None % narrow Roof P2 97 broad None P2 98 broad None P2 99 narrow None P2 100 narrow Roof PZ 101 narrow None PZ lm h d Raof P2 103 broad Rmf P2 Iû4 m w None N4 1 h d None N4 2 bmad Rnof N4 N4 3 nanow Roof 5 nanow RDof N4 6 brd Rmf N4 7 bruad RDof N4 10 narrow Roof N4 11 narrow Roof N4 12 nmw R d N4 13 nmw Roof 1- low 1- low 1- bw 1-1- hgh 1- tr* 1- low 1-1- low tugh 1- bw 1-1- knv 1- Low 1- bw 1- hgh 1- low 1- hlgh 1- hgh 1-2+ bw 1-1- hgh 1- medjum 1-1- bw 1- bw 1- bw 1- bw 1- kw 1- bw 1- Low 1- bw 1-1- bw bwes knuer d u m UPP medjum Iowa "PF lower U P F UPpcr UPPa U P F un= "PPCf U P F UPP- upper upper "P= narrow Roof 2+ mecüum upper ' nredum mqdwn bw meàum low bw bw mdwn bw bw d u m low bw Lnedum d u m nredum mrdum bw Iow UPper U P P UPper U P W UPper "Pper UPper U P F U P F UPper UPper "PF U P F U P F upi'er UPper U P P "PF "PP U P P U P F aedum lllodum 10wer Lowe lomr Iowe!r n\cdum bwa bwa mtdum m+dum lowa lowa Melane Masssam - h ge gull ptedatm on bùzck-kgged kttwrtkts 105

123 P NS L W R W D CP A N4 65 narrow None 2+ bw 1 N4 66 narrow None N4 67 narrow None N1 68 narrow Roof N4 70 now None N4 76 narrow None bw bw bw bw bw N4 77 braad None bw N4 80 narrow Raof lav N4 81 narrow R d N4 82 narrow Roof N4 83 narrow None N4 84 narrow None N4 86 mow Roof N4 87 narrow None N4 88 broad N4 89 broad None None N4 90 narrow None N4 91 broad N4 92 broad N4 94 broad N4 95 broad None Roof Roof None N4 % narrow None N4 97 narrow None N4 98 broad N4 99 broad N4 100 broad Roof None None N4 101 b d None N4 102 braad R d N4 l(x3 broad Roof N4 104 braad N4 105 bruad Roof None N4 106 narrow None N4 107 broad N4 108 broad Roof Roof N4 109 narrow Roof N4 110 namw Roof N4 111 narmw RDof N4 113 narrow Roof N4 114 narrow Roof hgh bw Low bw low low hgh d u m low low low low bw Cow bw P NS LW R W D CP A N4 115 nurowrd 1- I 116 brod Rd 117 hd W 118 nurow None 119 nurow Rad 120 METOW bd la nurow bd 122 nurow w la rwrow None 12s narrow None 126 narrow None 127 narrow None 128 nurow None I29 m w Roof 130 narrow None 131 nanrow None 132 m w W 133 narrow W 134 broad kd 136 narrow None 137 now None 138 broad None 140 brod None 141 narrow Roat 142 narrow W 143 bnrd None 144 ~lrow None 115 brcd R d 146 murow k d 147 br#d Rpd 148 brod Roof 150 kod M n brord M 152 bnrd Roof 153 hmd None 154 bmd None I56 breed None 1% brobd Roat 159 brad R d upper lomr Eowa lower UPF upper lower UPPW upper upper Melane Masssmo - Lmge gull pndatun on black-legged kttwakes 104

124 P NS LW R W D CP A N4 161 brmd R d 1- low upper 1 N4 162 bmd None low N4 165 narrow None low UPP" N4 166braad R d Iow U P F N4 167 brd Roof d u m N4 168 broad R d bw N4 169 broad RoDf N4 170 narrow Roof N4 171 broad Roof N4 IR h d Rod N4 173 broad None N4 174 narrow None N4 175 bmd Roof N4 176 broed None N4 177 nanow Roof N4 178 narrow Roof N4 179 narrow None N4 180 broad N4 181 bruad N4 182 broad Rad Roof Roof N4 183 b d Roof N4 184 bmad N4 185 broed N4 186 broad N4 187 bmad N4 188 bmad Roof Roof Roof Roof Roof N4 189 braad Roof N4 190 narrow R d N4 191 narrow R d N4 192 narrow Rmf N4 193 narrow Roof N4 194 m w N4 195 broad Roof None N4 1% M None N4 197 mow Rad N4 198 m w None N4 199 rarrow None PY4 200 narrow None N4 2fX narrow Roof Low lnedum low bw bw low lav Inu meâum DlQdum Eow bw Low low d u m bw lrmr UPP= U P F bwa bwa bwet bwa P NS LW R W D CP A N4 202 nurow Ibcd 1- ower 2m runow Rxd lower 204 m w bd Lnedum an nuow None lowa 206 nnow Rod 2.07 m w Rad twtow Nono m m w None 210 Mnow R d 211 bpd Rxd 212 runow Nom 213 nanow None 214 m w None 215 narrow None 2l6 narrow Nono 217 m w None 218 nnow None a9 m w R d m nurow R d m nan~w ~ o d 2z m w ROd 2z m w Rod 224 mmw None 225 bt#d kd 2%b#d Roa Whmd w 228 brad None 229 nurow None 230 mrmw Rod 231 ~ n o w Rrd a2 urrow None 233 nurow None 234 narrow None 235 nrrow None m6broed R d P7 brod R d 238 brnd None a9 bnrd None 240 bod None 4

125 f NS L W R W D CP A N4 241 brd Nom 1- hgh * N4 242 broad None N4 243 h d None N4 244 broad N4 265 bnad Rmf None N4 246 brwad R d N4 247 m o w R d N4 248 nanow Roof N4 249 narrow Roof N4 250 ~now Roof N4 251 narrow Roof N4 252 narrow Rmf N4 253 m o w None N4 254 mow Roof N4 255 nanow Roof N4 256 nanow Rmf N4 257 narrow None N4 258 narrow None N4 259 narrow b f N4 260 narrow Roof N4 261 m o w Roof N4 262 narrow Raf N4 263 m w Roof N4 264 narrow Rmf N4 265 narrow Roof N4 266 ~ n o w Roof N4 267 narrow Roof N4 268 narrow Roof N4 269 narrow None N4 270 h d None ~4 m brmd M N4 272 brmd None N4 273 m w Roof N4 274 broad None N4 275 h d None N4 276 narrow Roof N4 277 narrow Roof N4 278 narrow Rd N4 279 narrow Roof dm bwa d u m lower lower b w bwa 281 nurow Rod 282 namw Nom 283 r\~rrow R d 284 nurow Roa 285 nun>w Nom 286 ~ n o w None 207 nurow Rd 288 rkmw Roaf 289 mmow R d 290 Mnaw Rod 291 m w Rdu 292 nurow R d 293 nanow Rod 294 m w Rad 295 nurow Rod 2% nurow Roof 247 Mnow Ra4 296 Mnow Rod 299 m w Roa 300 m w R d 301 MnOw Rad 302 nurow kd 3a3brnsd Rmt 304 namw None 305blod Rwt mbrod Rad 3mlXuad Roof Roa ~ w Rwf 310 W kd 311 bpd None 312 numw Roof 313 nurow R d 314 numw Rod 315 narrow R d 316 broad Rmt 317 brad None 318 bod Roof 1- bw 1- bw 2+ lm? bw 2+ tow 1- bw 2+ low 1- Imv 2+ bw 1- kw 1- bjw 1- Iow 1- low 1- Low 1- tow 2+ lav 2+ ow r- low 1- Low 1- law 1- low 1- low 1- low 1- bw 1-2+ dm 1- hgh 2+ bw 1- bw 1- bw 1- bw 2+ hgh 1- low 1- bw lawer lomr lowa bwet lawer lower Lowa lower lower louer lower d u m lower meduln lower lower lower Iowa lower!ower Lower lowa Lower lowa lower lower

126 N4 339 narrow None N4 340 wow Roof N4 341 h d None N1 342 broad N4 343 braad None None N4 344 nmow Roof N4 365 broad N4 347 brd N4 348 broed N4 350 brmd None None Roof None N4 351 ~ now None N4 352 brcmd Rmf 55 1 narrow R d ss s5 2 nanow Roof 3narrowRoof rurow None SS S5 S5 S5 SS 5 narrow None 6 narrow None 7 namw Fkd 8 nanow Roof 9 oarrow None SS IO h d None SS Il bcctad None h@' kw hgh low low low bw mr?d~~ll UPP'= h& h& ksh hw kw bw low 1 N4 319 narrow None 1- bw bwa 1 1s 12 narmw Rm 2+ kw upf * NQ 321 narrow Rad bw lowe!? 13 bpd Nom 1- h N4 322 brd bf 14 rumw Nonc 2+ bw UPper N4 323 broad bof d u m 15 nurow Rd 1- bw maum N4 324 nanow RDaf d u m 16 nurow R d 2+ knv N4 325 broad R d 17 m w bd 2+ bw N4 326 brd Roof hgh 18 nurow Rod 1- bw N4 32ï b d Rad kw 19 narrow None 1- bw N4 328 namw None bw n) nurow Rmt 1- kw mqdum N4 331 narrow Roof fav bwa a -OW M 1- bw N4 332 btaad R d bw lower 22 llmow Rd 1- hw N4 333 narrow R#>f P Mrrow kaof 1- low N4 335 narrow None bw U P F 24 narrow None 1- low U P P N4 336 broad None low 25 ~ n o w Rad 1- U P P N1 337 namw None 26 narrow None 1- bw U P F N1 338 bmd None h& UPP= 27 mrtow RDQt 2+ U P F upper d u m upp= UPper "PP UPper UPper U P F UPPet U P F "PF "Pper UPper 20 Mnow Rod 29 narrow Rod 30 narrow RorJ 31 m w Rd 32 bmd RDQt 33 brod None 34 nurow None 35 brd None 36 k#d None 37 nurow R d 38 ~ n o w Rod 39 rmrmw R d 10 nurow Rad 41 nurow W 42 rurmw bd 43 nurowrod 44 nurow None 45 m w Rmt 46 numw Roof 47nurowRool 48 urrow W 49 nurow None 50 runow Rod h& 1- hgh 1- bw 1-1- hgh 1- lqw 1- hgh hgh 1- hgh 1- h@ 1-1- hgh 1- hgh 1- hgh 2+ h& 1-1- low 1-1- bw 1- hgh U P P "PF U P F lowa Lower d u m mtdom ~ U I n mrdun medun Lndum U P F

127 52 narrow Rad 53 narrow Roof 54 namow None 55 narrow None 56 narrow None 57 bmd None 58 nanow None "PPQ "Pper U P F U P F UPper UPP'" None bd None None None None None 59 m w 60 m w None None bwa None None 61 bruad None None 62 broad None None 63 b d None klwer ROd 64 broad None lower RuIf 65 brmd None 66 brmd None 67 broad None 68 broed None 69 h d None lower bwa Iower None w M Rod Rod 70 braad None 7ï broad None Iower Ralf None 72 broad Roof None 73 braad Roof None 74 braad None 75 braad None None None 76 brcad Roof 77 brmd Roof dm bw 78 braed R d 79 biaad Ro 80 braad Roof 81 brmd None 82 nanow Roof llcedum 83 broad Roof 84 broad None 85 narrow None 86 narmw Roof 87 brmd Roof 88 broad None 89 brnad None

128 P NS LW R W D CP A 5S 129 bruad None 1- bw narrow None bw 131 broad None 132 narrow Roof 133 narrow None 134 narrow None 135 narrow None 136 narrow None 137 m w 138 m w None None 139 narrow None 14û narrow None 141 bmd None 142 braad None 143 broad None 144 namw None 145 nanow Roof 146 narrow None 147 narrow None 143 mow None 149 narrow Roof 150 bmad None 151 broad None 152 broad None 153 narrow None 154 narrow None 155 narrow None 156 narrow None 157 narrow None 158 mow None 159 narrow None 160 narrow None 161 nnrrow None 162 h d None 163 broad None 164 broad None 165 broad None 166 h d None 167 broad None bw bw bw bw bw kw bw bw bw bw law hl kw low bw bow Low cnedum meàum Kgh h@' h& lower Iowa lower rnedjutn dm UPPer wum P NS LW R W D CP A S6 168 nvrow None 1- lé9 nuow None 170 hd None low n brcrd N- bw "Pl"= n aolld NO= 173 ~ n o w None 174 brad None 175 broad None 176 broad None 177 brod None 178 hmd None 179 bknd None IBO bnad None 181 brord None 182 brosd None 183 broad None 184 ûmed None 18s broed Rad 186 broed None 187 rrarrow None 188 m w None 189 ~arrow None 190 nurow R d 191 natrow Root 192 bnad Rad 193 brcd W 194 ~ m w None 195 Mm>w Rwt 1% Mnow 197 Mrrow Rod lm nurow Root 199 narrow Roa 200 MROW R d 201 nurow None 2û2 narrow None a)3 narrow R d 204 narrow Rwt 2m nurow Rod n]d nurow Rad h@ d u m h* w h@' h& bw w low low bw law meùwn bw kw law kw!au bw kw bw U P F U P P U P F lower lawa n\cdum rnedmm medtllll Melane Masssaro - Lmge gull prcdaton on bfack-legged ùtfurpkes 109

129 208 nanow Roof n39 narrow Roof 210 narrow w n ~ O W ~oof nzbtoed Roof 213 broad None a4 narrow Roof 2l5 nanow None 216 narrow Roof 217 narrow Roof 218 mmw R d 219 M~OW ~oor 220 narrow Roof m narr~w ~ oot 222 nanow b f 223 m w Roof 224 narrow None 225 broad Roof 226 broad Roof 228 ùrcad Roof 229 broad Roof 230 bmad Roof 231 broad Roof 232 h d Roof 233 broad Roof 234bd M 23s narrow RDof 236 h d Roof 237 broad Roof 238 broed None 239 narrow Roof 240 nanow Roof 241 narrow None 283 mmw None 244 b d None 245 h d None 246 broad None 247 bmad None I 1 $ 248 bmd NO~C 1- h@ bwa 1 lowa lower 10- bwa Iowa Iowa lower lowa ber bwa bwa bwa bwa 1- hgh 1- bu 1- bw 1- bw 1- bw Low 1- bw 1- bow 1- Low 1- bw bw 2+ bw 1- hgh 1- hgh 1-1- low 1- bw 1- low 1- kw 1- hgh 1-1- low 1- bw 1- low 1- hgh 2+ bw 1- hgh 1- bw 2+ lrnv 1- hgh 2+ kw 1- lava klwer Lower lowa lower "PF "PP "PF "PF U P F *PF Lnedum meduut upf "PF lower U P F Melane Masssmo - Large gull p'cdatcm on back-legged ktfwakes 110

130 P NS LW R W D CP A S5 289 narrow None 1- bw bwa b d None 291 namw None 292 narrow None 293 broad None 294 m w None 295 b d None 301 nanow None 302 hac None 303 narrqw Roof 304 narrow None 305 narrow None 3û6 narrow None 311 brvad None "PF UPper *PF lower UPper lower Melune Mnsssmo - Large gull prràatan on bladr-egged kttwaks 111

131

132

133 Meune Masssaro - Lmge gull predaton un btack-legged kttulokes 114

134

135 Melme Masssmo - LRge gull pndnta a black-kggtd kttwakps 116

136 Melane Masssmo - Lmge gull predaton on bhck-legged kttwakes 117

137 Melane Masssuro - Large gu1 prednton on blpck-kgged kîtwukes 118

138 Plot N4 (C)

139 Y) m * a (y O Melane Masssaro - Large gull pdaton on bùzck-legged khakes 120

140 Melane Masssmo - Large pl1 predatm on btadc-legged kttwakes 121

141 Plot

142 Plot Melane Masssmo - Large guli predaton on bkk-egged ktturakes 123

143

144