CausesandConsequencesof TerritoryChangeandBreedingDispersal
DistanceintheBlackKite
ManuelaG.Forero;JoseA.Donazar;JulioBlas;FernandoHiraldo
Ecology,Vol.80,No.4(Jun.,1999),1298-1310.
StableURL:
EcologyiscurrentlypublishedbyTheEcologicalSocietyof America.
YouruseoftheJSTORarchiveindicatesyour acceptanceofJSTOR’sTermsand Conditionsof Use,availableat Conditionsof Use provides,inpart,thatunlessyouhaveobtainedpriorpermission,youmaynotdownloadanentireissueofajournalormultiplecopiesofarticles, andyoumayusecontentintheJSTORarchiveonlyforyourpersonal,non-commercialuse.
Pleasecontactthepublisherregardinganyfurther useofthiswork.Publishercontactinformationmaybeobtainedat
EachcopyofanypartofaJSTORtransmissionmustcontain thesamecopyrightnoticethatappearsonthescreenor printedpageofsuchtransmission.
JSTORisanindependentnot-for-profitorganizationdedicatedtocreatingandpreservinga digitalarchiveof scholarlyjournals.FormoreinformationregardingJSTOR,.
FriDec3011:51:312005
Ecology,80(4),1999,pp. 1298 1310
© 1999by the EcologicalSocietyof America
CAUSES ANDCONSEQUENCESOFTERRITORYCHANGE AND BREEDING DISPERSALDISTANCE INTHEBLACK KITE
MANUELAG.FORERO,’JosÉ A.DONAZAR,JULIOBLAS,ANDFERNANDOHIRALDO
Department ofApplied Biology, Estación Biológica deDoñana, C.S.1.C.,Avda M Luisa s.n.,41013 Sevilla, Spain
Abstract. Factors affecting individualvariation inbetween-yearterritory change and dispersal distance were measured inalong-termstudy of Black Kites (Milvusmigrans) living inanarea of —100000 hainandaround Doñana National Park insouthern Spain. Adult birds(N =210) andfledglings (N =3061) wereindividuallymarked, andbreeding birds were monitored annually during 1989—1996todetect banded individuals.Of these
3271 birds, 652 breeding birds wereresighted,and164ofthem weresexed through cop
ulatory behavior. Nests werechecked annually tomonitor breeding performance.
Frequenciesof return between years were 83.1% for breeding males and 89.5% for breeding females.Frequenciesof territory change were 25.6% and 32.8%, respectively.
Females >8 yroldrarely changed territories.Dispersingbirdsdeparted significantlymore fromlow-qualityterritories(evaluatedthroughbreedingsuccessvariables).Breedingfailure andmate loss (divorce ordeath) favored breeding dispersion,bothinmales andfemales, but females changed territoriesmore frequentlyafter nest predation.Wedid not detect effects of resource availability,conspecific breeding density, or mate quality on the fre quency ofterritory change. Whenallthevariables wereincluded intoaGeneralizedLinear Model(GLM),onlybreedingsuccess, mateloss,andtheinteraction betweenthetwohad significant effects onterritory change. Dispersal distance ofbirdsthathadchanged territory wassimilar formalesandfemales (median 302m).Ninety percent ofindividualsmoved
<5km,andnonemoved >10 km.Distance moveddepended onthebird’sageandterritory quality, butaGEM model retained only ageasasignificant variable: older birds (males
andfemales) dispersedshorter distances.
These resultsindicate thatbreeding territory change anddispersal distance aredifferent individualdecisionsandaredeterminedbydifferent selectivepressures. Dispersal conse
quences weresignificant onlyforfemales: theytended tochange toterritorieswithalower risk of predation and obtained higher breeding success in the new location. Moreover,
individualsinthenewterritorieshadhigher breeding success thanthose occupyingaban doned territoriesthesame year.Dispersaldistances werenotcorrelatedwith thenewter ritory quality andprobabilityofbreeding failure. BlackKitesremained closetotheirformer breeding places;individual competitiveabilities maydetermine thefinaldispersal distance. Short-distancemovementsindispersal areprobably related tobenefits ofphilopatry(eco logical andgenetic) anddispersal costsassociated withfindinganewterritory inasaturated population,such astheonestudied.
Keywords: age and sexpatterns indispersal;Black Kite;breeding dispersal;breeding perfor
mance; dispersal distance; mate change; Milvus migrans; Spain; territory fidelity.
INTRODUCTION
Processesofdispersal invertebrateshaveimportant implications forpopulationstructureanddynamics(for reviews, seeGreenwood1980,GreenwoodandHarvey
1982, Johnson and Gaines 1990). Breeding dispersal is defined as movement of individualsbetween suc cessive breeding places and influences individual fit ness bydeterminingsurvival,mating success andre productive rates(Clutton-Brock 1988,Newton1989, Wiklund 1995,Waser1996).Inspiteofitsimportance, little is known about dispersal,mainly because gath eringlongitudinalinformationonindividualsiscostly,
Manuscriptreceived25June1997;revised 16May 1998;
accepted 9June1998.
E-mail:
andthespatialscaleof dispersalisoftentoolargeto relocate dispersedindividuals.
Sexandagearepotentiallyimportant sourcesofvari
ation in dispersal probability(Newton and Marquiss
1982, Part andGustafsson1989). Within asexorage class, individualvariation indispersal patterns maybe relevant (seereferencesinWaser1996,Wiklund1996). Most studies haveattempted toidentify theproximate factors affecting breeding dispersal.A frequent finding from awide range of vertebratespecies is that indi vidualsmaychangeterritory following apoororun successfulreproductiveattempt ormate loss (see ref erences in Nager et al. 1996, Waser 1996, Wiklund
1996), or after occupying a poor quality territory (BoyceandBoyce1988,Montalvo andPotti1992,Kor pimaki1993).Acommon shortcomingofthesestudies,
1298
however,isthedifficultyofdisentangling therelative importanceof each factor, as they frequently covary (e.g., Goodburn 1991, Choudhury1995).
Theadaptive significanceofvertebratebreedingdis persalhasreceived littleattention,duetomethodolog icalconstraintsimposed bylong-term studies. Some authors havefound thatbreeding dispersal clearly did notresult inanaccrual ofindividualfitness (Newton and Marquiss 1982, Pusey and Packer 1987, Rood
1990, Smale etal.1993, Olsson etal.1997). Inthese cases, breeding dispersal couldbeinterpretedasanon adaptiveor neutralprocess(PayneandPayne1993). However,breedingdispersalhas alsobeenshownto enhance fitness(Shields 1987,BoyceandBoyce1988, PartandGustafsson1989,Boutinetal.1993,Stenseth andLidicker1992,Payneand Payne1993,Part1995, Waser1996).
Individual decisions regarding breedingdispersal should beregulated byatrade-off between manypos siblecostsandbenefits(Greenwood1980). Benefitsof philopatryin breeding dispersal processes can bere lated tosuchecologicaland/or genetic componentsas sitefamiliarityanddominance,gene-localadaptations andkin-selectionpatterns (e.g., increasedtolerance of relatives infeeding incursions,reduced escalationof boundary disputes; see Hinde 1956, Maynard Smith
1964,Clutton-BrockandHarvey 1976,Greenwood and
Harvey 1976, Greenwood1980, GreenwoodandHar vey1982, PartandGustafsson1989,Beletsky andOr ians1991).Onthe otherhand,potential advantagesof breeding dispersal include gaining higher quality ter ritories or mates and improvingsurvival probability, future breeding performance;hence, individualfitness (Part andGustafsson1989;seereviewformammals in Waser 1996). The costs of breeding dispersal should berelated toahigher riskofmortality (seereviewfor birdsinFleischer [19831,Eden[19871,andStrickland
1991; andfor mammals inWauters etal. [1994] and
Waser 1996) and thedifficulty of finding agood ter ritory(Greenwood1980,Askenmo 1984,PartandGus tafsson 1989, Bensch andHasselquist1991). Dispers ing birds and mammals endure higher predation risk andenergy expenditurewhenthemovementsarelong (Gaines and McClenaghan1980, Johnson andGaines
1990,Waser1996).Themagnitudeofthesecosts,how
ever,woulddepend onaspecies’ lifehistory patterns. Inthis paper, westudy causes andconsequencesof breeding dispersal intheBlack Kite(Milvus migrans), a medium-sizedmigratory raptor, in which dispersal occurs onaspatial scale thatwecanmeasure. Forthis purpose, wehaveusedinformationfrom an8-yrmon itoring program of banded individuals.Wehave ana lyzed theeffects ofsexandage,local density ofcon specifics, territory and mate quality, breeding experi ence,andbreeding performance.Theroleofeachvari able was examined univariately. We then used GeneralizedLinear Models (GLM) toassess thevari anceexplained byeachvariable andreducecovariance.
AtlanticOcean
o510 1520
•kms
•Villages
— Roads
BlackKite
::Breeding Areas
Marshes
BiologicalReserveLimit
——
+NationalParkLimit
FIG. 1. Studyarea.Limitsofprotectedareas areshown. ZonesofmaximumBlackKitebreeding densityareinside theBiologicalReserveandinthenorthernpartoftheNational Park.
Moststudiesof breedingdispersalconsideronlyone variable todefinedispersal patterns:dispersal distance, whichcanbezero(nodispersal).Inthisstudy,however, weconsideredthatindividuals couldmaketwoinde pendent decisionsduring adispersal process: whether ornottochange territories,andhowfartomove(dis persal distance).For this reason, weanalyzed site fi delity anddispersal distance separately.Ourmainob jectives were (1) toidentify proximate factors deter mining territory change, (2) todetect ifthese factors affect breeding dispersal distance,and(3)toexamine theconsequencesofterritory change anddispersal dis tance.
METHODS
Study area
Thestudy wascarried outinDofianaNational Park (insouthwesternSpain) andsurroundingareas(atotal of100000ha;6°12’—6°40’W,36°48’—37°20’N)(Fig.
1).The three mainhabitats inthepark areseasonally drying marshland,Mediterraneanscrublandwithscat teredcorkoaksQuercus suber,andcoastal sanddunes with stone pines Pinus pinea. Amosaic ofcultivated lands andpine plantationsoccupies surroundingareas with trees along the banks of streams. The average monthly temperature duringspring andsummer ranges from 13.7°C inMarch to23.2°C in August. Average
annual rainfall is —540mm,concentratedinautumn— winter (September—March) and is variable among years.During themainstudyperiod(1981—1996),pre cipitationranged252—1032mm.Formoredetailed de scriptionsofthearea,seeValverde(1958) andRogers andMyers (1980).
The species
TheBlack Kiteisamigratory,medium-sizedbirdof prey (630—941g;Cramp andSimmons 1980) present inthestudyareaduringMarch—August.Itnestsintrees, mainly cork oaks, pines and eucalyptus,usually bor dering marshy areas (Fig. 1).Males provision females andnestlingsin earlystages,andmostincubationand brooding iscarried out byfemales (Cramp and Sim mons 1980; F Hiraldo, personal observation). The breeding populationwasstableduringthestudyperiod, oscillating around 550 breeding pairs. Nonbreeding birds typically reach a total of 400—500individuals annually, concentratinginsixcommunal roosts. Young birds return totheir natalareatobreed(M.0.Forero, J.A.Donazar,andFHiraldo, unpublished manuscript).
Field procedures
The monitoringof this population (completecen
suses and metal banding ofsome nestlings)began in
1964attheBiologicalReserve, acore areaintheNa tionalParkof 7600ha(Fig.1).From1973on,metal banding wasextended toallnestlings inthisarea.Dur ing1986—1996allthenestlings inthiszonewerealso marked withpolyvinyl chloride (PVC)bands,atanage of20—35d.EachPVCbandhadanalphanumericcode with three characterslegible upto300 m,with a20—
120X telescope.In1986, wealsobegan tobandnest lings(metalandPVCbands)intherestofthebreeding areas.Additionally,since 1986wecaptured adultbirds (1yrold)withcannonnetsandpaddedlegholdtraps, throughoutthestudyarea.Someofthesebirdswereof known age, asthey hadbeen banded withmetal band as chicks in previous years (since 1964). A total of
2123 nestlings were marked only with metal bands, while3061nestlings and210adultbirds weremarked withmetalandPVCbands.Webegantolookforband edbreeding birds intheReserve areain1989; during
1991—1996,we extended this activity to the whole
studyarea.During 1992—1996,wealsoregularly mon itored six communal roosts to look for nonbreeding banded birds; twodays aweek these were monitored fromnoonuntildusk. Priorto1996, werecorded 2035 marked individuals:652inbreeding territories,808in roosts,and767individualsofunknown breedingstatus (feeding oncarrion, perching onatree,orfounddead). From1992onwards, wesexed164breedingindividuals (87malesand77females) throughcopulatorybehavior recorded between early March andlate April.
Allbreeding pairsinthestudy areaweremonitored
every year to detect banded birds. Moreover, in the
BiologicalReserve wevisited allneststocheckbreed-
ing parameters;intherest ofthestudy area, weonly visited thosenestsinwhichatleastonememberofthe pair carried a PVC band. Each monitored nest was checkedatleastthreetimesthroughout thebreeding period,recordingclutchsize,hatching, and fledging success. Weestimatedlaying date from alinear re gression relating eighth primary length toage(Vinuela andBustamante1992). Werecorded eggorchick pre dationwhenwefoundclearsignsofpredator presence: fresh eggs ornestlings partially eaten andclaw marks inbranches,ornestbowlsdisturbed.Themostcommon species depredating nestsofBlackKitesinthispop ulationareLynx (Lynxpardina)andGenette(Genetta genetta),and,infrequently,theSpanish Imperial Eagle (Aquila adalberti).Thetwomammalianpredators have nocturnalhabits andcan potentiallycatch adult birds onthenests.
Because fieldstudies ofdispersal areoftenbasedon finitestudyareas, manydispersal distributionsforver tebrates arebiased (seereview inMoore andDolbeer
1989, Koenig et al. 1996). Inour study, such abias shouldbelowforthefollowing reasons: (1)Return frequenciesin our population are high (73.5%) and similar tosurvival estimations(73.7%; M.G.Forero, unpublished data)calculated through “Jolly—Seber” capture—recapture methods (Lebreton et al. 1992), whichclearly indicates thatincomplete sampling of returning birdsisnotapparent(Martinet al.1995). Moreover, capture—resightingprobabilities calculated through thesamemethodologywereveryhigh(94.7%; M.G.Forero, unpublished data).(2)We intensively searched forbanded birdsatlargerdistances ( 40km) thanthelongestdispersal distance (10km)thatwehave detected; (>90%ofdispersalswerewithin aradius of
5 km)(Fig.1).(3) Therehavebeennometalband recoveriesof adult birds during the breeding period outside thestudy area (N =50 recoveriesfrom birds banded only withmetal).
Territory limits andquality
Territorial boundarieswereobtained byplotting the position of the nesting trees on 1:10000maps. We defineaterritory asthe3-haareadefended byabreed ingpair(BustamanteandHiraldo 1993;FHiraldo,per sonal observation).Thisareawasmeasured bytracing a100mradius circle around the nesting treethefirst timeitwasdetected. These territoriesweresubdivided when two different pairs occupied two nesting trees insidethedefinedcircleinthesameyear.Inthesecases, therestofthetreeswithinthe3-hacirclewereincluded inone or the other territory,through the behavior of thebreeding pairs:copulations,courtship,defense, and perches ofadults andyoung.
Often, territory quality has been measured through
twodifferent kindsofvariables:thoserelated tophys ical habitat characteristics(e.g., Tye 1992), andthose related to breeding performance(e.g., Montalvo and Potti 1992). The Black Kiteisasemicolonialspecies
TABLE 1. Variables used toanalyze thefactors determiningterritory change and dispersal distancesinBlack Kites.
Individualcharacteristics
Age(yr): Known ageinyears.
Sex: Determinedviaobservationofcopulation.
Ageofprevious mate(yr): For theeffect ofthisvariable,weonlyconsideredthoseindividualswhoseformer mates were present inthefollowingbreeding season. Thus, weavoided mixing theeffects ofmatedeath andageofprevious mate.
Breeding experience(yr): Weconsideredthe sequenceof breeding attempts only in birds for which the first breeding
attempt was known forcertain.
Previousbreedingconditions
Predation:Predation (1)ornopredation (0)ofeggs orchicks.
Breeding success: Successful(1,atleast one chick fledged) orunsuccessful(0).
Mate change: Mate change ornomatechange, from one yeartothenext.Three categories:0,nochange; 1,widowhood,
and 2,divorce.
Environmentalcharacteristics
Local density:Number ofother occupiedterritorieswithin 1-km radius.
Winter rainfall: Total volume (L)rained before thebreeding season (September—March).
Territoryquality
Percentageofyears with predation:Yearswith nest predation aspercentageofthetotal number ofbreeding attempts. Productivity:Total number ofchicks fledged divided bythetotal number ofbreeding attempts.
Productivitywithout predation:Asfor theformer variable, excludingyears inwhich predation occurred.
Percentageofyearsofoccupation:Yearsinwhichtheterritory wasoccupied divided bythetotalnumberofyearsmonitored.
Fledgingratet:Mean number offledglings raised peryearduring successfulbreeding attempts.
Percentageof years with breeding successt: Successfulyears (1 young fledged) divided by thetotal number of years
monitored.
t Variables excluded from further analysis onthebasis ofaPCA (see MethodsandTable 2).
with communalforagingareas(Valverde1967,Veiga andHiraldo1990);however,breedingsuccessin this speciesisstrongly influenced byfoodavailabilitywith intheir small defended territories(Vinuela andVeiga
1992,Vinuelaetal.1994).Thus,wechosesixbreeding performance variables toquantify territory quality (Ta ble1).These variables evaluate different components ofterritory quality,suchasnestattractiveness,resource availability,individualability, andnestpredation risk. As some of these variables probably covary, we re ducedtheirnumberthrougha PrincipalComponent Analysis (PCA) with VARIMAX rotation (Pielou
1984), onthebasis of84territories,afterlog-normal
ization ofthevariables.We obtained twoaxeswith eigenvalues>1thataccounted for69.4% ofthevari ance(Table2).FactorIclearly separated thevariables. Percentageof years of occupationand percentage of years with predation were isolated from each other around the coordinateorigin. On the other hand, we found the other four variables to be associated with
eachother:productivitywithoutpredation,productiv ity, fledgingrate,andpercentageofyearswithbreeding success. Basedontheseresults, weeliminatedthelast twovariables forfurther analyses.
Breeding sitelocations and distance measurements
Latitude andlongitudewererecorded foreach nest usingaGlobalPositioning System (GarminPersonal Navigator).Weonly considered those measurements with anerror <25 m. Breeding site fidelity and dis persal distances (straight-linedistance in meters be tween previous and present year’s nest site) were re stricted to those birds with at least two successive breeding records.
Statistical analyses
First, we analyzed the data univariatelyfollowing SiegelandCastellan (1988).Second, weusedGener alized Linear Models (GLM) (Nelder andWedderburn
1972, Dobson 1983, McCullaghand Nelder 1983) to
TABLE2. Loading factors forthePrincipal ComponentAnalysis onlog-normalizedvariables measuringterritory quality (N
=84territories).
Variable
Percentageofyears with predation Productivitywithout predation Productivity
Percentageofyears ofoccupation
Fledging rate
Percentageofyears with breeding success
Eigenvalue Percent variance Cumulative variance
Factor I
—0.320
0.916***
0.967***
0.332
0.741***
0.732***
3.076
51.22
51.22
Factor II
0.705***
0.117
—0.062
0.697***
0.115
—0.280
1.090
18.21
69.43
***
Highly significant(P0.001) correlationsbetween factors andvariables.
deriveamathematicaldescriptionof(1)theprobability ofbreeding territory change, and(2)thedispersal dis tance. Wefitted the explanatoryvariables to the ob served datathroughamodification ofthetraditional forward step-wiseprocedure (seedetails inDonázar et al.1993).GUMsoftware (BakerandNelder1978)was used.
Fortheanalyses, we chosevariables evaluatingin dividual andenvironmental characteristics,territory quality, andother previous breeding conditions(Table
1).Yearcandeterminechanges inenvironmentalcon
ditions andinfluences breedingsuccess (Clutton-Brock
1988, Part 1991). In our study area, foraging Black Kitespreferentially exploitmarshareas.Thishabitat undergoes strong interannualvariation inwaterlevels and, consequently,infood resources(Valverde 1960,
1967). Therefore,weused previous winter rainfall as an indicator of changes in resource availability for Black Kites(Valverde 1958,VinuelaandSunyer 1992, F.Hiraldo, personal observation).
The analyses were carried out separatelyfor males andfemales.Asinformation onbreeding successand other variables was lacking for a few birds, sample sizes varied somewhat between analyses.
RESULTS
Individual variation indispersal performance
Priortostatisticaltreatments,weanalyzed individual variation in dispersal patterns.Wetried to detect if birdsthathadchanged territoryafterthefirstbreeding record had ahigher probabilityof dispersal between thesecond andthird attempt thannondispersingbirds afterthefirstbreeding record.Theresults showed that, for birds that hadchanged after thefirstseason (N
17),41.1%changed afterthenextbreeding season. For birdsthatdidnotchangebetween thefirsttwobreeding seasons (N =56),thepercentagewas35.7%. Thedif ference was not significant (Fisher’s exact test, P
0.777).
Individual variation in dispersal distance was ex amined through arepeatabilityanalysis followingLes sellsandBoag(1987). Repeatabilitiesofdispersal dis tancewere—0.087(F56=0.836, P=0.568)formales, and0.472 (F5,7 =3.000, P=0.092) for females.Al thoughfemale valuescanbeconsiderated“moderate” (Harper 1994), lack ofstatisticalsignificancepermits usto assumethatdispersaldistanceisnotstrongly influenced byindividual factors. Thus,alltheseresults permitus totreateachbirdandsitechangeasanin dependent unit(Wiklund 1996).
Factors affecting territory change
Sexandage.—Returnfrequenciestothestudy area for PVC-banded breeding birds were 83.1% of 142 males and89.5% of143females.ForallPVC-banded breeding individuals (males, females, and unsexed birds, N=419) thefrequencywas73.5%. Thehigher
FIG.2. Frequenciesofterritorychangeinrelationtoage formalesandfemales. Thenumbersabovethebarsrepresent samplesizeofthatageandsex.
returnrateobtained whenincludingbirdsofknownsex couldresultfromthehigherprobabilityofsexingthose birds living longer.
Territory change between successivebreeding sea Sonsshowed similar frequenciesin male and female Black Kites (25.6% of 121 vs. 32.8% of 122, x2
1.182, df=1,P=0.277). Forboth sexes combined,
frequency ofterritory changewas29.6%of 284birds. Therewasnosignificant correlationbetweenfrequency ofterritory changeandageofindividuals formales, considering 11 classes of known age birds (r
—0.117, P=0.776,N=9).Forfemales thecorrelation
wassignificant (r —0.828,P= 0.008, N=9),but was influenced by low values inoldest birds (>8 yr old;Fig.2). Afterremovingtheseoldestfemales,the correlationwasnotsignificant(r=—0.429,P=0.419, N = 6).
Effects ofterritoryquality and breeding success.— Wechose thefour variables defined bythePCA(see Methods: Territorylimitsandquality, andTables1and
2) to test the effects of territory quality on territory changefrequencies.Formalesandfemales,productiv ity and productivity without predation were signifi cantlylowerinterritories thatwereabandoned.For females, asimilar trendwasdetected regarding theper centageofyearswithpredation,andformalesin the percentageofyears ofoccupation(seeTable3).
Thelocaldensity ofconspecificsandresourceavail abilitycouldbeimportantfactorsdeterminingterritory change: changes would beless probable inareas with high breeding density and good food foraging condi
TABLE3. Attributesofindividualsandterritoriesintheyear beforeaterritorychangeoccurredvs.otherbirdsthatdidnot
change.
ChangedterritoryDidnotchangeterritory
Variable / Sex / Mean / SD / N / Mean / SD / N / PAge ofpreviousmate(yr) / M / 6.25 / 1.71 / 4 / 5.79 / 1.96 / 18 / 0.67
F / 5.67 / 3.05 / 3 / 6.17 / 2.21 / 12 / 0.83
Breeding experience(yr) / M / 2.77 / 1.01 / 13 / 2.98 / 1.37 / 50 / 0.79
F / 2.81 / 1.18 / 27 / 3.31 / 1.40 / 42 / 0.16
Local density (pairs) / M / 27.23 / 21.28 / 30 / 21.38 / 13.05 / 38 / 0.71
F / 20.22 / 13.71 / 37 / 22.34 / 14.84 / 74 / 0.34
Percentageofyears with predation / M / 0.101 / 0.134 / 24 / 0.122 / 0.135 / 89 / 0.45
F / 0.195 / 0.195 / 36 / 0.092 / 0.120 / 81 / 0.002
Productivity / M / 0.422 / 0.347 / 24 / 0.715 / 0.430 / 89 / 0.02
F / 0.484 / 0.500 / 36 / 0.781 / 0.457 / 81 / 0.003
Productivityexcludingpredation / M / 0.474 / 0.433 / 24 / 0.798 / 0.482 / 87 / 0.002
F / 0.581 / 0.580 / 36 / 0.865 / 0.472 / 79 / 0.01
Percentageofyears ofoccupation / M / 0.530 / 0.189 / 19 / 0.701 / 0.211 / 67 / 0.001
F / 0.546 / 0.213 / 29 / 0.658 / 0.263 / 59 / 0.06
Notes: SD,standard deviation;N,sample size; andP,probability.StatisticalcomparisonsweremadewithaMann-Whitney
Utest.
tM,male; F female.
tions (e.g., Wikiund 1996). The local density (1 km around thenest) variedamong areasintherange1—72 pairs (mean =22.26, SD =15.29, N=250). There werenosignificant differencesinlocaldensity ofcon- specificsbetweenindividuals (malesandfemales) changing territory,andthosebeingfaithful toit(Table
3).Previous winterrainfall,asapredictor offoodavail ability inthefollowingspring, had noeffect onterri torychangeforeithersex:correlationsbetweenwinter rainfall andannual frequencyofterritory change were notsignificant (males, r=0.036, P=0.963, N=7; females, r=0.456, P=0.30, N=7).
Numberofyearsbreeding (breedingexperience)has
beenproposedto explainterritory lossinbirds(e.g., Jakobsson 1988).Thishypothesis predicts thatthe probability ofterritory changewoulddecreaseasthe breeding experienceofindividualsincreases.Wefound noeffect ofthisvariable: breeding experiencedidnot differ between individualsthatdidanddidnotchange territories (Table 3), but very old females rarely changed territories(Fig. 2).
Breeding failure due to nest predation and other causesmaypotentiallydetermineterritory change(e.g., Sonerud 1985). Weexamined whether ornotcomplete breeding failure influenced the reoccupation of the same territory in the next breeding season. Seven of
54 successfulmales (13%) and seven of 52 females
(13.5%) changed territories. Forunsuccessfulbirds, frequencieswere37.1% of62males and47.6% of63 females. Forbothsexesdifferencesweresignificant (males, x2=6.685, df=1,P= 0.010; females, x2=
13.706, df=1,P0.001). Inaddition, weexamined
thesingle effect ofnestpredation onterritory change thefollowingseason. After anest predation event, 3 of 17 males (17.6%) and 13 of 23 females (56.5%)
changed territory.Inthose nests that were notdepre dated, frequencieswere27.3% of99malesand26.1% of92females.Thedifferencesweresignificant forfe
males (x2=6.478, df =1, P=0.011), but not for
males (x2=0.289, df =1,P=0.591).
Todetect whether nestpredation presented ariskfor breeding birds, we examined return rates during the season following nest predation. Eighty percent of males(16of20males) whose nests werepredated re turned to the study area the following season. When thenestwasnotpredated, return rate was83.6% (102 of122 males; Fisher’s exact test, P=0.742). Forfe males the rates were 84.0% (21 of 25 females) vs.
90.7% (107 of 118 females; Fisher’s exact test, P
0.230). These results suggest that adults are rarely killed during predation events.
Effectsofmatequality(age).—Lowmatequalityand
mateloss(through divorce ordeath)couldincrease the probabilityof territory change (e.g., Wiklund 1996). Formaleandfemale Black Kites,ageofprevious mate (as anindicatorofquality) had noeffect onterritory change (Table3).On theother hand, mateloss hada strong influence for females. When afemale lost her mateshechanged territory witha higherfrequencythan whensheretainedhim(78.6%of14vs.40.0%of 35; Fisher’s exact test, P=0.025). Formales, wefound a similar trend (69.2% of 13 vs. 37.1% of 35; Fisher’s exact test, P=0.055).