ANSER ANSER IN VEJLERNE, DENMARK

137 EGG PREDATION IN REEDBED NESTING GREYLAG GEESE ANSER ANSER IN VEJLERNE, DENMARK JENS NYELAND KRISTIANSENl,2 Kristiansn J.N. 1998. Egg prdation in rdbd nsting Grylag Gs Ansr ansr in Vjlrn, Dnmark. Arda 86: 137-145. Egg prdation in Grylag Gs Ansr ansr nsting in rdbds was studid at thr sits in Vjlrn (NW Dnmark). Ovrall, 34% of all clutchs hatchd succssfully (rang 26-53%), whil at last 34% of all nsts wr abandond. Avian prdation of clutchs was particularly high on on sit (20%) compard to othr sits (0-2%), probably as a rsult ofbttr brding opportunitis for crows in that ara. Mammalian gg prdation was low in th two sits with natural barrirs in th form of watr channls or with highr vgtation dnsitis. Nsts nighbouring dprdatd nsts wr mor likly to b dprdatd. An important factor affcting gg prdation by mammals was th distanc of th nst to rd cutting tracks. Scond, thr was a tndncy for succssful nsts to occur in aras with highr vgtation dnsitis. Mammalian prdators wr Rd Fox Vulps vulps and Amrican Mink Mustla vison and/or Wstrn Polcat M. putorius. Ky words: Ansr ansr - gg prdation - rdbd - mammalian prdators lth National Environmntal Rsarch Institut, Dpartmnt of Coastal Zon Ecology, Kal, Grnavj 12, DK-8410 R nd, Dnmark; 2Univrsity of Copnhagn, Zoological Institut, Dpartmnt of Population Biology, Univrsittsparkn 15, DK-2100 Copnhagn, Dnmark, -mail: JNKristiansn@zi.ku.dk INTRODUCTION Th rproductiv output of gs varis gratly within and btwn spcis (Johnson t al. 1992). On of th most important factors dtrmining nsting succss is prdation (Campbll 1990; Sargant & Ravling 1992) and a varity of avian and mammalian prdators may tak ggs and young, dpnding on nst sit and gographical location. Som natural sits such as islands offr protction from prdators, and island nsting birds may suffr lowr prdation than lswhr (Giroux 1981; Wright & Gils 1988; Lokmon & Woodward 1992). Prdation may also b rducd in nsts concald by vgtation (Witkowski 1983; Johnson t al. 1992) or whr accss to nsts is rstrictd to prdators by othr barrirs lik channls or moats (Lokmon & Woodward 1993). Grylag Gs Ansr ansr brding on islands and in rdbds in Dnmark prsumably slct nst sits so as to rduc thir accssibility to mammalian prdators, or hid thir nsts in dns vgtation to avoid dtction by avian prdators. This study xamind th importanc of factors affcting gg prdation from Grylag Goos nsts in rdbds, and focussd on th nsts' accssibility to mammalian prdators, as masurd by vgtation dnsity surrounding th nst sit, watr lvl at th nst sit, distanc of nsts to land and rd cutting tracks, and nst dnsity. STUDY AREA Th study was carrid out in th 6000 ha natur rsrv of Vjlm in NW Dnmark (Fig. 1). Study sits wr slctd at Slbjrg/Glombak Vjlr (SGV, 504 ha), Bygholm Nord R0rskov (BNR, 368 ha) and T0mmrby Fjord (TF, 161 ha). Vjlm rsrv is an important wtland ara for brding and staging watrfowl in Dnmark, and consists of larg shallow frshwatr laks,

138 ARDEA 86(2), 1998 Fig. 1. Map of Vj1m. Aras shadd dark dnot rdbds. L0N =L\'lnnrup Fjord, 0ST =0stri1d Vj1, V-A = Vs1\'ls/Arup Vj1, TF =T\'lmmrby Fjord, LUN = Lund Fjord, HAN = Han Vj1, SGV = S1bjrg/Glombak Vjl, BNR =Bygho1m Nord R\'lrskov. 1500 ha of rdbds, and xtnsiv madows (Mpllr 1980). Th rdbds ar dominatd by Phragmits australis, with smallr patchs of Typha angustifolia, T. latifolia, bulrush Scirpus spp. and locally also willow scrub Salix cinra and S. aurita and hawthorn Cratagus spp. Som 700 pairs of Grylag Gs brd at Vjlrn, almost xclusivly in th rdbds. Th rdbds of th Slbjrg/Glombak Vjl and Bygholm Nord Rprskov study sits ar xtnsiv and wid and in Bygholm Nord Rprskov ar, in som placs, surroundd by watr channls. In Tpmmrby Fjord th rdbd is narrow and surroundd by watr channls and dns vgtation of willow and hawthorn. Rd cutting is carrid out on a rgular basis in all thr aras and cutting of smallr patchs was don in th wintr prcding th study laving ths patchs, and a numbr of tracks mad by th rd cutting machin xposd. METHODS On th basis of vrtical arial photographs takn on 12 April 1994 from an altitud of 300 mtrs, 136 nsts wr locatd and subsquntly visitd aftr th incubation priod in 1994. Nar ach nst, four 50 cm x 50 cm frams wr placd North, South, Wst and East of th nst cntr. Within th fram all rd stms standing mor than 75 cm abov th substrat surfac wr countd. 20 wr chosn randomly and thir diamtr was masurd (in mm) using a digital callipr at 60 cm from th substrat surfac. Th fram was positiond sufficintly far (usually 1.5 2 m) from th nst to avoid aras of rros damagd by gs. For ach nst an avrag rd stm diamtr (d = 2r) and rd stm dnsity (D) was calculatd and ths wr usd to provid an ara indx of th vgtation covr, mm 2 /m 2 (%0 co-

Kristiansn: EGG PREDATION IN GREYLAG GEESE 139 vr), rprsnting ach nst (assuming rd stms to b circular, vgtation covr indx =n r 2 D). Th fw nsts found in rd-mac wr tratd in th sam way to rds. Only old rd stms wr masurd, so as to rflct th conditions th gs facd during nst building and incubation, whn nw rd shoots had not mrgd. Watr lvls wr masurd in ach of th four frams and th man dpth adjustd by using valus obtaind from prmannt stations (whr watr lvls wr masurd at last onc vry fiv days all yar), to obtain valus rprsnting th conditions whn th brding gs wr choosing nst sit (c. first wk of March). A man watr lvl valu was thn calculatd for ach nst. Nst fat was dtrmind as a succssful hatch, prdation by mammals or prdation by birds (Rardn 1951; Bang & Dahlstrom 1980; Brown t al. 1992). Prdation was assumd to hav occurrd if thr wr signs of prdator prsnc associatd with th nst. Th typ of prdator was idntifid from th way th gg was tratd, th typ of facs, th prsnc of hair or othr signs and tracks (Rardn 1951; Bang & Dahlstrom 1980; Brown t ai. 1992). Usually thr was mor than on sign of th prdator (.g. both facs, hair and gg shlls). Nsts wr rcordd as succssful (ggs hatchd) if thr wr no signs of prdator prsnc and if th rmains of at last on gg showd signs of having hatchd succssfully. Nsts without gg rmains but which wr clarly from th prsnt yar (nwly built or with down prsnt) wr rcordd as dsrtd. Arial vrtical photographs wr usd to masur th distancs of nsts to land and to rd cutting tracks; nst dnsity (nsts ha- 1 ) at intrvals of 100 or 200 m from land at two sits of Slbjrg/Glombak Vj1 (SGVA and SGV B) and on sit in Bygholm Nord R rskov (aras wr masurd by ali. 3100 Ara mtr, Li-Cor, Inc. Lincoln, Nbraska, USA); and th distancs btwn all nsts at on sit in Bygholm Nord R rskov and two in Slbjrg/Glombak Vj1 to assss th ffct of nst dnsity on gg prdation (s statistical sction, Randomisation tst). In ach of ths sits goos nsts showd a non-random clumpd distribution. Statistical analyss wr prformd by mans of SAS statistical packags (SAS Institut Inc. 1990). Diffrncs btwn man valus of variabls wr tstd with unpaird Studnt's t-tst or unpaird Wlch-approximat t-tst. Stpwis logistic rgrssion analysis (PROC CATMOD) was usd to stimat th ovrall ffct of variabls on th probability of mammalian prdation in ach of th thr locations. Nst fat (prdation/succssful hatch) was chosn as th dpndnt variabl and vgtation covr, watr lvl, distanc to land and distanc to rd cutting track as th indpndnt variabls. Discriminant function analysis was carrid out to show how many of th nsts dprdatd by mammals, and how many of th succssful nsts, wr classifid corrctly according to th abov variabls. Fischr's Exact tst was prformd to tst if th narst nighbour to a dprdatd nst usually was also dprdatd. Randomisation tsts (Mont Carlo procdur; Sokal & Rohlf 1981) wr prformd to dtrmin whthr actual distancs btwn prdatd nsts wr qual to or lss than distancs btwn a random st (100 000) of hypothtical prdatd nsts. X 2 tsts or Fischr's Exact tsts wr prformd to tst for diffrncs btwn sits in hatching succss, mammalian and bird gg prdation and nst dsrtion. Sparman rank corrlation was mad to tst for corrlation btwn nst dnsity and distanc from land and for prdation rat and distanc from rd cutting tracks. All tsts wr two-taild. RESULTS Sit diffrncs, nst dnsity and distanc from land Ovrall 34% of nsts wr succssfully hatchd ranging from 26% in T mmrby Fjord (TF) to 53% in Bygholm Nord R rskov (BNR; Tabl 1). Egg prdation was 22% (15% by mammals and 7% by birds). Of all nsts 34% wr dsrtd but th dgr of dsrtion was not signifi-

..._-~---~---"'- 140 ARDEA 86(2), 1998 Tabl 1. Fat of 136 Grylag nsts distributd in Slbjrg/Glombak Vjl (SGV), Bygholm Nord R rskov (BNR) and T mmrby Fjord (TF). Locality Prdation Prdation Succss Abandond Uncrtain Total (birds) (mammals) (dsrtd) fat SGV 1 2% 15 26% 18 31% 21 36% 3 5% 58 BNR 0 0% 2 6% 17 53% 11 34% 2 6% 32 TF 9 20% 3 7% 12 26% 14 30% 8 17% 46 Total 10 7% 20 15% 47 34% 46 34% 13 10% 136 5 4 ~3..c (f) iii ~2 o 200 400 600 distanc from land (m) "'BNR OSGVB SGVA 800 Fig. 2. Nst dnsity (nsts ha- 1 ) and distanc from land (m) in Slbjrg/Glombak Vjl A (SGV A; filld circls), SGV B (opn circls) and Bygholm Nord R rskov (BNR; filld triangls). cantly diffrnt btwn localitis (X 2 2 = 0.3875, P = 0.8238). Bird prdation was significantly mor frqunt in T0mmrby Fjord (20%; X 2 2 = 15.308, P < 0.001) compard to th othr aras (Slbjrg/Glombak Vjl 2% and Bygholm Nord R0rskov 0% rspctivly). Mammalian prdation was significantly lowr in both Bygholm Nord R0rskov (6%) and T0mmrby Fjord (7%) than in Slbjrg/Glombak Vjl (SGV; 26%; X 2 2 = 10.609, P < 0.01). Although 80% ofall gs nstd in pur rd stands (n = 136), 19% of nsts wr build in a mixtur of rds and rd-mac. No significant diffrncs btwn th proportion of nsts prdatd by mammals in th two main vgtation typs P. australis and P. australis/typha sp. wr dtctd in any of th thr aras: Slbjrg/Glombak Vjl (Fischr's Exact tst, Odds ratio =1.065, n =58, P =1), Bygholm Nord R0rskov (Odds ratio =1.078, n =32, P = 1) and T0mmrby Fjord (Odds ratio = 0.2632, n = 46, P = 0.3491). In rdbds of on part of Slbjrg/Glombak Vjl (SGV A; Fig. 2) nst dnsity incrasd with distanc from land (r = 1, n = 6, P < 0.01). In anothr part (SGV B) and in Bygholm Nord R0rskov th incras was not a significant (SGV B r = 0.4, n =4, P =0.75; Bygholm Nord R0rskov r = 0.4, n =4, P =0.75). Howvr, this was probably du to vry small sampl sizs. It was not possibl to carry out this analysis in T0mmrby Fjord du to th narrow width of th rdbd in this ara. Factors affcting mammalian gg prdation Dtails of nsts suffring mammalian prdation ar prsntd intabl 2. Although man vgtation covr valus wr highr for succssful nsts in all localitis, th diffrnc was only significant in Bygholm Nord R0rskov. No diffrnc in man watr lvls btwn dprdatd and succssful nsts wr found in any of th thr aras. Th distanc from land to nst location was significantly lss for succssful nsts than dprdatd nsts in Slbjrg/Glombak Vjl and a similar (non-significant) tndncy was found in Bygholm Nord R0rskov (no data from T0mmrby Fjord). Th distanc of prdatd nsts from tracks lft by rd cutting machins was lss than that of succssful nsts (data only from Slbjrg/Glombak Vjl; r =-0.95, n =4, P =0.0833; Fig. 3).

._-~.-- ---~ Kristiansn: EGG PREDATION IN GREYLAG GEESE 141 Tabl 2. Man valus (± SD) for vgtation covr, watr lvl, distanc to land and distanc to rd cutting tracks for succssful nsts and nsts dprdatd by mammals for ach study sit. Slbjrg/Glombak Vjl Prdatd Succssful man±sd man±sd t-valu n df P vgtation covr (mm 2 /m 2 ) 2905.6 ± 1508.1 3521.1 ± 1960.5 0.9788 33 31 n.s. %0 covr 2.9 %0 3:5 %0 watr lvl (cm) 33.8 ± 7.7 30.8 ± 7.8 0.9511 33 31 n.s. distanc to land (m) 146.0 ± 90.31 63.7 ± 80.5 2.237 22 20 <0.05 distanc to track (m) 28.4 ± 21.1 129.5 ± 72.9 5.316 31 17 <0.001.-.-----~.,.._- Bygholm Nord R rskov vgtation covr (mm 2 /m 2 ) 2106.7 ± 863.6 3266.1 ± 755.8 2.409 20 18 <0.05 %0 covr 2.1%0 3.3 %0 watr lvl (cm) 19.5 ± 5.5 22.6 ±4.5 1.052 18 16 n.s. distanc to land (m) 26.7 ± 32.1 14.5 ± 6.8 0.6549 20 2 n.s. T mmrby Fjord vgtation covr (mm 2 /m 2 ) 4605.2 ± 550.8 5634.4 ± 1125.2 1.508 15 13 n.s. %0 covr 4.6%0 5.6%0 watr lvl (cm) 39.5 ± 4.3 40.6 ± 4.1 0.4259 15 13 n.s. 100 80 60 ~ '"0 (J.) 0.. 40 n (J.) c 20 o 79% n=14 0% 0% n=4 n=7 o 50 100 150 200 250 distanc from tracks (m) Fig. 3. Rat of prdation (%) with incrasing distanc from rd cutting tracks. Th logistic rgrssion (Tabl 3) tstd th null hypothsis that nst succss probability was indpndnt of: vgtation covr, watr lvl, distanc to land and distanc to rd cutting tracks. Th tst was prformd sparatly for ach locality. Th only significant variabl ntring th modl was distanc to rd cutting tracks in Slbjrg/Glombak Vjl which xplaind 77% of th variation. From this it can b prdictd that th prdation risk dclins with incrasing distanc to ths tracks. Discriminant functional analysis corrctly classifid 96% of all 23 nsts in Slbjrg/Glombak Vjl with known fat incorporating both distanc to land and distanc to rd cutting tracks in th modl. Only on nst (a succssfully hatchd) was misclassifid (Fig. 4). A dprdatd nst was usually situatd clos to anothr d prdatd nst and a succssful nst to anothr succssful nst (Fischr's Exact tst, Odds ratio =13.619, n =45, P < 0.001). In ach of th aras that wr tstd with th Randomisa-

142 ARDEA 86(2), 1998 Tabl 3. Logistic rgrssion (SAS, CATMOD procdur) for factors affcting nst prdation of Grylag Gs. Each locality was tratd sparatly. Only th variabl that ntrd th modl is prsntd (distanc to track in Slbjrg/Glombak Vjl). Th maximum liklihood (ML) asssss how wll th modl fits. Variabl Estimat X 2 P ML Tabl 4. Randomisation tst (Mont Carlo procdur). Numbr of computd distancs (runs): 100 000 btwn a random st of hypothtical dprdatd nsts compard to th actual distancs btwn dprdatd nsts in Bygholm Nord Rl'lrskov (BNR) and Slbjrg/Glombak Vjl (SGV A and B). P is th probability of obtaining a distanc lss than or qual to th actual distanc btwn prdatd nsts. constant 2.7905 7.07 0.0078 0.767 Locality Numbr of Numbr of P-valu distanc to track -0.0500 6.42 0.0113 nsts prdatd nsts BNR 16 2 0.017 SGVA 8 5 0.035 SGVB 12 9 0.0445 300 ~250.s -g 200.!!1 ~ 150 0 Ql g 100 ll 0 n '0 50 0 o o o prdatd nsts succssful nsts 0,I----"----r-----r--_r _----'r--<_---r-----,--j o 50 100 150 200 250 300 distanc from track (m) Fig. 4. Distanc from land (DL) and tracks (DT) of succssful (filld circls) and dprdatd (opn circls) nsts, with th discriminant lin (0.0336DL = 0.0498DT - 1.2143). tion tst, th actual distancs btwn dprdatd nsts wr significantly lowr than th distancs of a numbr (100 000) of randomly chosn hypothtical dprdatd nsts (Tabl 4). Hnc nighbouring nsts usually had a similar fat; spcially if thy wr situatd clos togthr. Prdators Hoodd Crow Corvus coron comix wr th only potntial avian gg prdator on goos ggs in th ara, although non wr actually obsrvd ating ggs during this study. Crows frquntly nstd in trs and bushs in th vicinity of th r- dbds and wr oftn sn on prmannt lvatd look-out posts. Unfortunatly no quantitativ information is availabl concrning th prsnc of this spcis. On fiv occasions facs of Rd Fox Vulps vulps wr found in goos nsts and thr tims clos to nsts. Som facs rvald rmains of gg shlls and chorion mmbran. Foot prints wr rcordd thr tims. Foxs wr obsrvd svral tims during th fild work both on th madows and in th tracks of th rd cutting machins. On four occasions facs of Amrican Mink Mustla vison or Wstrn Polcat M. putorius wr found in nsts, and foot prints blonging to on of ths prdators wr sn twic in th tracks of th rd cutting machin. DISCUSSION Hatching succss smd quit low (c. 34%) in comparison with othr studis on brding Grylag Gs;.g. Nwton & Krbs (1974) 42-70%, Witkowski (1983) 55%, and Nilsson & Prsson (1994) 58-78%. Howvr, th rsults of th prsnt study could asily b undr-stimats in such a comparison. In this study, nsts which wr partly prdatd but still may hav yildd succssfully hatchd ggs wr assignd as prdatd (i.. th prdator had found th nst), contrary to

Kristiansn: EGG PREDATION IN GREYLAG GEESE 143 most othr studis whr such nsts wr assignd as succssfully hatchd (Nwton & Krbs 1974; Witkowski 1983; Nilsson & Prsson 1994). Hnc, comparisons with othr studis should b tratd with caution. Th major causs ofhatching failur wr nst dsrtion (34%) and prdation (birds and mammals combind; 22%). Nst dsrtion was rlativly high compard to othr studis (Nwton & Krbs 1974; Witkowski 1983; Nilsson & Prsson 1994). Th causs of nst dsrtion ar unknown but disturbanc by prdators or humans may hav bn th caus (MacInns & Misra 1972; Witkowski 1983). In this study, som nst dsrtion could, howvr, rsult, according to my dfinition (nsts without signs of shlls), from a prdator rmoving all ggs from a nst. Bcaus no othr potntial avian gg prdators wr rsidnt in th ara, th main avian prdator was almost crtainly crows, spcially around T mmrby Fjord, whr prdation rat was highst. Potntial crow brding sits sm to covr th ntir priphry of th rdbd of T mmrby Fjord and th rdbd forms only a rlativly narrow strip (<500 m in dpth), unlik th xtnsiv rdbds in both ofth othr aras. Hnc, it would hav bn much asir for a bird prdator to locat nsts from its look-out post in T mmrby Fjord than lswhr. Erikstad t al. (1982) found mor prdation by crows on Willow Ptarmigan Lagopus lagopus nsts within 350 m of thir own nsts than on nsts furthr away. Bird prdation on goos nsts implis that th incubating goos is absnt from its nst sinc it is unlikly that a crow is capabl of chasing th goos off th nst. In both T mmrby Fjord and Bygholm Nord R rskov, prdation by mammals was much lowr than in Slbjrg/Glombak Vjl. Accssibility to th rdbd in T mmrby Fjord was rathr difficult if a mammal has to approach from th ground. This sit had th highst vgtation dnsity and thr ar only two points of accss without swimming for a mammal across th channl which surrounds th ara. Th nsts in Bygholm Nord R rskov could only b rachd via a 2.5 km long, narrow dik without any scap possibilitis or through I km of dns rdbd from th ast. Furthrmor, in Bygholm small channl had to b crossd. Of th thr aras, Slbjrg/Glombak Vjl offrs th asist accss for a mammalian prdator. Rd dnsity is rathr low and no channl surrounds th rdbd. Hnc, trrstrial prdators can ntr from almost any point along th priphry of th rdbd. This may xplain th high rat ofmammalian gg prdation thr. Only in Bygholm Nord R rskov was thr a significant diffrnc btwn vgtation covr around nsts dprdatd by mammals and succssful nsts, but thr was th sam (non-significant) tndncy in th othr aras. This suggsts that Grylag Gs nsting in dnsr rdbds aras ar bttr protctd against mammalian prdators than ar thos nsting in mor opn rdbds. Watr lvl at th nst sit did not affct gg prdation, probably bcaus watr lvls wr too low (rang 20-40 cm dpth) to show any ffct. Foxs ar occasionally sn in aras surroundd by narrow channls and such watr barrirs wr unlikly to hav dtrrd th mor aquatic-adaptd Amrican Mink or Wstrn Polcat (Nithammr & Krapp 1993) from ntring th rdbd. Givn knowldg of high dnsitis of pry in an ara occupid by brding gs or othr spcis of birds, a mammalian prdator (vn a Rd Fox) may attmpt crossing watr barrirs to gt accss to th brding ara. Th highr and somwhat unxpctd succss rat of nsts situatd clos to land could b du to th fact that nst dnsity tndd to incras with distanc from land (Fig. 2). Nsts clos to land wr lss dnsly distributd and, as discussd blow, gg prdation (which incrass with nst dnsity) is strongly rlatd to th fat of narby nsts and th distanc btwn ths. Ths factors, combind with th ngativ corrlation btwn nst prdation and distanc to tracks, suggst that mammalian prdators avoid ntring rdbds dirctly from land but choos th mor accssibl tracks which lad thm to aras of mor abundant food rsourcs (Fig. 5). As wll as concalmnt and th dg ffct, gg prdation was rlatd to th fat of narby nsts. Th risk of dtction incrasd markdly if

144 ARDEA 86(2),1998 ACKNOWLEDGEMENTS Fig. S. Gnralisd illustration of a hypothtical ara ofrdbds with rd cutting tracks and Grylag Goos nsts in rlation to land ara. Th figur shows why it may b mor profitabl for a mammalian prdator to follow tracks cut by th rd cutting machin which ar likly to lad thm to highr nst dnsitis than thos ncountrd during similar sarch ffort clos to land (whr nst dnsitis ar lowst). Filld circls indicat goos nsts; hypothtical rd cutting tracks ar shown. a nighbouring nst was dprdatd. A prdator that has found a nst in an ara, will probably hav stayd and sarchd for mor nsts, or rturn latr (Stphns & Krbs 1986; O'Rilly & Hannon 1988). Sinc nsts in ths aras showd a clumpd distribution, prdation risk might hav bn rducd if nsts had bn mor vnly spacd (Sugdn & Byrsbrgn 1986). Goos activity in a rdbd ara is frquntly rvald by indistinct paths of brokn rds mad by th gs thmslvs. As th frquncy of ths paths probably incrasd with nst dnsity, aras of high dnsitis may b asir to locat for th prdator. Th paths may also mak th rdbd mor accssibl. In summary, gg prdation by mammals on nsts of Grylag Gs at Vjlrn wr affctd by dg ffct (distanc from nst to rd cutting tracks), nst dnsity (fat of, and distanc from, narby nsts) and possibl also by nst concalmnt (vgtation) and prdator accssibility du to physical barrirs (channls). I am gratful to th Aag V. Jnsns Foundations who partly financd this study and gav prmission to carry out th fildwork at th Vjlrn natur rsrv. I am indbtd to Lin Louis Slilrnsn for assistanc during th fildwork and commnts on th manuscript. Spcial thanks to Tony Fox for his advic and ncouragmnt during this study and for his commnts on th manuscript. Jan Dyck, Lif Nilsson and Jspr Madsn ar also thankd for advic and commnts on th manuscript. Spcial thanks to Gosta Nachman who kindly wrot th softwar program for th randomisation procdurs. Finally I thank Jlilrgn Ptr Kjldsn and th staff of th fild station of th Vjlrn natur rsrv for assistanc on diffrnt mattrs during th study. REFERENCES Bang P. & P. Dahlstrom 1980. Dyrspor. Gads Forlag, Copnhagn, Dnmark. Brown R, J. Frguson, M. Lawrnc & D. Ls 1992. Tracks and signs of th birds of Britain and Europ, an idntification guid. Christophr Hlm, A & C. Black. London. Campbll B.H. 1990. Factors affcting th nsting succss of Dusky Canada Gs Branta canadnsis occidntalis, on th Coppr Rivr Dlta, Alaska. Can. Fild-Nat. 104: 567-574. Erikstad K.E., R Blom. & S. Myrbrgt 1982. Trritorial Hoodd Crows as prdators on Willow Ptarmigan nsts. J. Wildl. Manag. 46: 109-114. Giroux J.E 1981. Us of artificial islands by nsting watrfowl in southastrn Albrta. J. Wildl. Manag. 45: 669-679. Johnson D.H., J.D. Nichols & M.D. Schwartz. 1992. Population dynamics of brding watrfowl. In: Batt B.D.J.,AD. Afton, M.G. Andrson, C.D. Ankny, D.H. Johnson, J.A Kadlc, & G.L. Krapn (ds) Ecology and managmnt of brding watrfowl: 446-487. Univrsity of Minnsota Prss, Minnapolis. Lokmon J.T. & RO. Woodward 1992. Nsting watrfowl and watr birds on natural islands in th Dakotas and Montana. Wildl. Soc. Bull. 20: 163-171. Lokmon J.T. & RO. Woodward 1993. An assssmnt of prdator barrirs and prdator control to nhanc duck nst succss on pninsulas. Wildl. Soc. Bull. 21: 275-282. MacInns C.D. & RK. Misra 1972. Prdation on Canada Goos nsts at McConnll rivr, Northwst Trritoris. J. Wildl. Manag. 36: 414-422.

Kristiansn: EGG PREDATION IN GREYLAG GEESE 145 M0llr H.S. 1980. Naturforholdn i Vjlm. Forskningsrapport nr. 17. Frdningsstyrlsn, Milj0ministrit, Dnmark. Nwton 1. & R.H. Krbs 1974. Brding of Grylag Gs Ansr ansr on th outr Hbrids, Scotland. J. Anim. Eco!. 43: 771-783. Nitharnmr J. & F. Krapp (ds.) 1993. Handbuch dr Siiugtir Europas, 5: 1-2. AULA-Vrlag, Wisbadn. Nilsson L. & Prsson H. 1994. Factors affcting th brding prformanc of a markd Grylag Goos Ansr ansr population in south Swdn. Wildfowl45: 33-48. O'Rilly P. & SJ. Hannon 1988. Prdation of simulatd Willow Ptarmigan nsts: th influnc of dnsity and covr on spatial and tmporal pattrns of prdation. Can. J. Zoo!. 67: 1263-1267. Rardn J. 1951. Idntification of watrfowl nst prdators. J. Wild!. Manag. 15: 386-395. Sargant A.B. & D.G. Ravling 1992. Mortality during th brding sason. In: Batt B.DJ., A.D. Afton, M.G. Andrson, C.D. Ankny, D.H. Johnson, J.A. Kadlc & G.L. Krapn (ds) Ecology and managmnt of brding watrfowl: 396-422. Univrsity of Minnsota Prss, Minnapolis. SAS 1990. SAS usr's guid: SAS/STAT, vrsion 6: 1 & 2. SAS Institut Inc. Sokal R.R. & F.J. Rolf 1981. Biomtry. 2nd dition, Frmann, Nw York. Stphns D.W. & J.R. Krbs 1986. Foraging thory. Princton Univrsity Prss, Princton. Sugdn L.G. & G.w. Byrsbrgn 1986. Effct of dnsity and concalmnt on Amrican Crow prdation of simulatd duck nsts. J. Wild!. Manag. 50: 9-14. Witkowski J. 1983. Population studis of th Grylag Goos Ansr ansr brding in th Barycz vally Poland. Acta Om. 31: 179-216. Wright R. & N. Gils 1988. Brding succss of Canada and Grylag Gs Branta canadnsis and Ansr ansr on gravl pits. Bird Study 35: 31-36. SAMENVATTING In ht natuurrsrvaat Vjlm in ht noordwstn van Dnmarkn wrd ondrzocht in hovrr d daar in ritlandn nstlnd Grauw Ganzn Ansr ansr t lijdn haddn van prdati van d irn door vogls n (land-) zoogdirn. Aangnomn wrd, dat d toganklijkhid van d ritlandn n d zichtbaarhid van d nstplaats blangrijk aspctn warn waardoor d kans op prdati van ht lgsl wrd bynvlod. Ht al dan nit uitkomn van d irn wrd afglid van sporn op n rond ht nst (ht uitrlijk van d irdoppn, sporn van prdatorn), tijdns n bzok kort nadat d ganzn mt hun jongn uit ht gbid vrdwnn warn. In ht rsrvaat wrdn d brodrsultatn n prdatikansn vrglkn in dri dlgbidn (Slbjrg/Glombak Vjl (SGV), Bygholm Nord R0rskov (BNR) n T0mmrby Fjord (TF); Fig. 1). Gmiddld kwam in ht ghl gbid 34% van d lgsls uit (spriding 26% in TF, 53% in BNR; Tabl 1), trwijl vnns 34% van d nstn voortijdig blk t zijn vrlatn. In T0mmrby Fjord wrd lifst 20% van d lgsls door vogls gprdrd, trwijl dit aandl in Slbjrg n Bygholm rspctivlijk 2% n 0% bdrog. D blangrijkst prdator was vrmodlijk d Bont Kraai Corvus coron cornix, di in T0mmrby Fjord vl gschikt nstglgnhid blk t hbbn. D prdati van lgsls door zoogdirn was laag in d tw gbidn waar grppls n slotn natuurlijk barrirs vormdn (6% in BNR, 7% in TF). In Slbjrg/Glombak Vjl wrd 26% van d lgsls door zoogdirn gprdrd. D blangrijkst prdatorn in Vjlm warn Vos Vulps vulps n Bunzing Mustla putorius of Amrikaans Nrts M. vison. Ht blk dat dz zoogdirn handig gbruik maaktn van 'padn' in d ritlandn di vroorzaakt wrdn door ritsnijdrs in ht gbid, waardoor zondr al t vl inspanning d brodplaatsn mt d hoogst dichthdn ganzn kondn wordn brikt. Aangtoond wrd dat n nst n vl grotr kans lip om door zoogdirn gplundrd t wordn wannr in d onmiddllijk omgving ook andr nstn bzocht warn. D mst succsvoll nstn wrdn aangtroffn op grot afstand van drglijk 'togangswgn' n in gbidn mt rlatif dicht vgtati. (CJC) Rcivd 20 Dcmbr 1996, accptd 12 March 1998 Corrsponding ditors: Lo Zwarts & Ks (Cl.) Camphuysn