TheAuk 116(2):355-363, 1999

REPRODUCTIVE SUCCESS AND NEST-SITE SELECTION IN A COOPERATIVE BREEDER: EFFECT OF EXPERIENCE AND A DIRECT BENEFIT OF HELPING

B. J. HATCHWELL,• A. E RUSSELL,M. K. FOWLIE,AND D. J. Ross Departmentof and Plant Sciences, University of Sheffield,Sheffield S10 2TN, UnitedKingdom

ABSTRACT.--Wedetermined whether nest-site characteristics influence reproductive suc- cessand whetherexperience influences nest-site selection in a populationof cooperatively breedingLong-tailed Tits (Aegithaloscaudatus). Nest predationwas high; only 17%of breed- ing attemptsresulted in fledgedyoung. The heightof nestswas an importantdeterminant of success;low nestswere significantlymore successfulthan high nests.A breeder'sage, natal nestsite, and breedingexperience had no significanteffect on nest-siteselection. How- ever,failed breeders who helped at thesuccessful nests of conspecificsbuilt subsequentnests lowerthan nestsbuilt prior to their helpingexperience. Failed breeders who did not help showedno reductionin tlseheight of subsequentnests. Moreover, the subsequentrepro- ductivesuccess of failed breederswho helped was significantlyhigher than that of failed breederswho did not help.We concludethat helpersgain informationon nest-sitequality throughtheir helping experience and thus gain a directfitness benefit from their cooperative behavior.We suggest that experience as a helperoffers a morereliable cue to nest-sitequality thanbreeding experience because helpers are associatedwith nestsonly during the nestling phasewhen few nestsare depredated.In contrast,although successful breeders may expe- riencesuccess with a low nest,they are evenmore likely to haveexperienced the failureof low nestsbecause of the high rate of nestpredation. Received 26 December1997,accepted 28 July1998.

A MAJORDETERMINANT of reproductivesuc- acteristics,then no consistentselection may ex- cessfor manyorganisms is the abilityof breed- ist for choiceof particularnest sites. However, ers to protecttheir offspringfrom predation. if breeding successis consistentlyrelated to Thisis particularlytrue of manyopen-nesting certainnest traits, then high ratesof nestpre- passerineswhere the rate of nestpredation may dation will exert strong directional selection be extremelyhigh (Lack 1954,Ricklefs 1969, againstthe choiceof low-qualitysites by breed- Martin 1995). A variety of antipredatorstrate- ers, resultingin low variancein the critical gieshas evolvedamong to reducenest characteristics.Alternatively, in the absenceof predation, including colonial or dispersed consistentselection for particular nest sites, breeding,use of cavities,nest concealmentor site choicemay be a variabletrait, with adap- camouflage,elaborate nest design, and protec- tive nest-sitechoice occurring either through a tive nesting associations(see Collias and Col- learningprocess of trial and error, or as a con- lias 1984).Numerous studies have shown an in- ditionalbehavior depending on predationrisk traspecificrelationship between nesting suc- or habitattype. cess and various characteristics of nests or nest Individual breeding performance can in- sites, e.g. nest structure (Baeyens1981), nest creasewith age and experiencein many ver- density (Anderssonand Wiklund 1978,Potts et tebrates(see Clutton-Brock 1988), and this has al. 1980, Hatchwell 1991, Chamberlain et al. often been attributedto increasedforaging ef- 1995, Meilvang et al. 1997), and nest conspic- ficiency(e.g. Desrochers 1992a, b) or increased uousness(Picman et al. 1993, Hatchwell et al. reproductiveeffort; i.e. the "constraint"and 1996),although other studieshave shownno "restraint"hypotheses of Curio (1983).Indi- sucheffect (e.g. Holway 1991, Colwell 1992, Fil- vidualsmay acquireforaging skills at anytime, liater et al. 1994,Cresswell 1997a). If predation but more specializedparenting skills, suchas risk is randomwith respectto nest-sitechar- nest placementor incubation,may only be ac- quiredthrough breeding. Such experience may be gainedthrough personal reproduction (e.g. E-mail: [email protected] Marzluff 1988, Marzluff and Balda 1992) or

355 356 HATCHWELLET AL. [Auk, Vol. 116 throughobservation of others(e.g. Boulinier et we investigatewhether and how individuals al. 1996), but among cooperativebreeders it acquire information about nest-sitequality. In also may be acquiredby helping a breeding particular,we examinethe hypothesisthat ex- pair rear its offspring(Brown 1987). Few stud- periencegained through helping at the nestof ieshave shown a directbenefit to helpersof in- conspecificsinfluences subsequent selection of creasedexperience, although notable excep- nest sites. tionsexist. In White-wingedChoughs (Corcor- ax rnelanorharnphos),helpers acquire foraging METHODS skillsduring their long prebreeding association with cooperativegroups (Heinsohn et al. 1988), We studieda populationof 15 to 35 pairsof Long- and in SeychellesWarblers (Acrocephalus sechel- tailed Tits from 1994 to 1997 in the Rivelin Valley, lensis),both helpersand breedersgain experi- Sheffield,United Kingdom. The study site (ca. 3 km2) compriseda variety of habitatsincluding hedge- ence in nest building, nest guarding, and in- rows, scrub, mature deciduous woodland, and small cubation (Komdeur 1996). standsof coniferoustrees. The breedingattempts of In this study,we used the cooperativebreed- individually markedbirds were closelymonitored ing systemof the Long-tailedTit (Aegithalosthroughout each breeding season(March to June). caudatus)to investigatethe effects of nestsite on We found a total of 178 nestsbelonging to 67 males reproductivesuccess and the effectsof experi- and 68 females.Each individual wasrepresented by ence on nest-siteselection. Long-tailed Tits betweenone and eightnests in our dataset. In some build domednests in very diversesites and suf- analyses,this raises the potentialdifficulty of pseu- doreplication,but we consideredeach nest to be in- fer high ratesof failure,largely caused by pre- dependentin analysesof reproductivesuccess with dation (Lack and Lack 1958,Gaston 1973, Glen respectto site characteristics.This is justified be- and Perrins1988). A new nestis built for every causealthough some nestsbelonged to the same breedingattempt, and a pair builds between birds, each was located in a different site, and it is oneand four nests in a singleseason. Each nest nest location that is important for this question constitutesa large reproductiveinvestment be- (Hatchwell et al. 1996, Cresswell1997b). Further- causethey are amongthe mostelaborate struc- more, nests of the same individual often encom- tures of any Europeanbird speciesand require passedthe wholerange of possiblenest sites, and the a lengthybut variablebuilding period (the av- same individuals did not build successive nests at the sameheight (males, F = 1.33,df = 38 and 97, r eragebuilding period for firstnests of the sea- = 0.086; females, F = 2.28, df = 36 and 81, r = 0.294). sonwas 25 days,although later nests were built In analysesconcerning experience and nestsites, we in just 8 days;B. J.Hatchwell unpubl. data). The used individuals as independentdata. Divorce was nestis made of mossbound with spider'ssilk, frequent both within and between seasons(B. J. the exterior is covered by several thousand Hatchwellunpubl. data), so we treatedthe sexes sep- piecesof lichen,and the interioris lined with aratelyin analysesof nest-siteselection. an averageof about 1,500 feathers(Hansell The greatmajority of nestswas found during the 1993,B. J.Hatchwell unpubl. data). The sexra- early stagesof building.Although we madepartic- tio is 1:1in the cooperativebreeding system of ular efforts to find replacementnests following nest failure, a few nests were never found; this was usu- Long-tailed Tits, and all birds attempt to re- ally attributableto renestingof failed pairs outside producein pairs.If their ownbreeding attempt the studyarea and sowas unlikely to biasthe sam- fails,however, breeders may become helpers at pling of nests.Nests were visited every one to three the nestsof their relatives,assisting in the care days during building, laying, and incubation and of nestlingsand fledglings (Lack and Lack were observedevery two days during the nestling 1958, Gaston 1973, Glen and Perrins 1988, periodto recordthe presence and identity of helpers. Hatchwell and Russell 1996). Thus, helpers Frequentnest visits might elevatenest predation maygain information about what constitutes a rates if potential predatorsobserve such visits or if successfulnest site through their own breeding visits reducenest concealment(Mayfield 1975, Len- experienceand/or by helpinganother pair. ington 1979).We do not considerthat an observeref- fect on nest predationbiased our resultsin any way First, we investigatewhether the high pre- for three reasons.First, depredatednests were usu- dation rate of Long-tailedTit nestsis influ- ally torn apart (see below), creating a "drift" of encedby characteristicsof the nest site. We feathersfrom the lining, so the survivalof the great show that nestplacement plays an important majorityof nestscould be checkedfrom a distanceof role in determiningbreeding success. Second, several meters. Nest contents of accessible nests were April 1999] Experienceand Nest-site Selection 357

checkedonly to confirm the start of laying, clutch 50 size,hatching, and during bandingof nestlings.Sec- ond, mostof the nestsplaced high in treeswere rel- 40' H [] Failed atively inaccessibleand so were visited only if they survived to day 11 of the nestling period, when

chickswere banded(laying, incubation,and hatch- c 30' ing datesfor thesenests were recordedby observa- tion of parentalbehavior). Therefore, if nestvisits in- creasedpredation rates, low nestswould havehad a E• 20 lower successrate than high nests,the oppositeof z the pattern found (seebelow). Third, evidencefrom an extensive experimental study (n = 766 nests) of 10 open-nestingpasserines in Britain found no effect of nest visits on predation rates (Mayer-Gross et al. 1997). 0 We recordedthe plant speciesthat providedthe 0 2 4 6 8 10 12 14 16 main supportfor the nest.Nests were found in 15 Nest height (m) plant species, mainly bramble (Rubusfruticosus; 23%),gorse (Ulex europaeus; 16%), holly (Ilexaquifol- FIG.1. Frequencydistribution of heightsat which ium; 15%),or hawthorn(Crataegus monogyna; 11%). Long-tailedTits built their nests(n = 158). Theremainder were in conifer,birch (Betula spp.), or alder (Alnus glutinosa)trees. Nest sites were classi- fied asbeing either in a tree forkor supportedby pe- mostlikely avian predators;in threeinstances ripheralbranches and alsoby the type of vegetation predatorswere seenat nests(two were jays and being either "protected" by thorns on stems or one a ).Other speciesuse active leaves(e.g. holly, bramble,gorse, etc.) or "unpro- Long-tailedTit nestsas a sourceof nest mate- tected" (mostlyother tree species,but alsohoney- rial and may havebeen responsible for a small suckle [Lonicerapericlymenum] and bracken [Pteri- proportion of abandonments;however, they diumaquilinum]). Nest height was measuredto the were not implicatedas a causeof nestingfail- nearest0.5 m; a simpletriangulation technique was ure onceeggs had been laid. Weasels(Mustela usedto measurethe height of neststhat couldnot be nivalis)and gray squirrels( Sciuruscarolinensis) reacheddirectly. Nest predatorswere identified from nestremains; were abundantin the studyarea and probably corvidstypically tear off the top of the nest,whereas were responsiblefor all depredationsby mam- mammalianpredators gain accessvia the nesthole mals. or enlarge the nest entrancein a less destructive The causeof nestingfailure variedthrough- manner than avian predators(Gaston 1973). Four- out the nesting cycle.Most of the adandon- teen nests that were surrounded with chicken wire ments(95%; n = 20) occurredduring thebuild- for protection againstpredators were included in an- ing phase,often when bad weatherdisrupted alysesof nest-siteselection because the nestswere early nests.Nest predation(n = 104) occurred initiatedbefore we protectedthem; these same nests during all phasesbut wasmost frequent during were excludedfrom analysesof breeding success. incubation(40%, vs. 15% during nestbuilding, Meansare given ñ 1 SD unlessotherwise indicated. 27% during laying,and 17% during the nest- ling phase;X2 = 14.96,df = 3, P < 0.01).Failure RESULTS during building occursbecause avian preda- tots cannot see a nest's contents without first Reproductivesuccess.--Only 17% (n = 158 to- removing its roof. Avian predatorswere most tal) of the nestsproduced fledglings. The main likely to destroynests during the laying and in- causeof breedingfailure was predation(79%) cubationphases (70% of cases;n = 81),whereas and abandonment(15%). The remaininglosses mammalianpredators were more likely to de- were causedby bad weather (2%), death of a stroy nestsduring the nestlingphase (50% of parent (1%), or unknown factors(3%). Nest re- cases;n = 14), when olfactory and auditory mains suggestedthat birds were the main cuesmay facilitatenest location. predators(85% of 95 nestswhere the likely Nest sites and reproductivesuccess.--Nest predatorswere identified), the remainder being heightvaried from 0.5 to 17 m (Fig. 1) and had mammals. EurasianJays ( glandarius) a significant effect on reproductivesuccess; and Black-billedMagpies (Picapica) were the 23% of the nestswithin 2.25 m of the ground(n 358 HATCHWELLET AL. [Auk, Vol. 116

TABLE1. Nestheight (f ñ SE,with numberof breedersin parentheses)in relationto ageof breedersfor all nestsbuilt by known-ageLong-tailed Tits in eachyear and the first nestbuilt by known-agebirds in each year.

Age of breedera Sex First year Secondyear Third year All nests Female 4.47 ñ 1.72 (5) 3.44 _+1.29 (3) -- Male 3.89 -+ 0.77 (25) 2.97 -+ 0.70 (13) 1.38 + 0.37 (2) Total 3.99 ñ 0.69 (30) 3.06 -+ 0.61 (16) 1.38 -+ 0.37 (2) First nests only Female 4.20 ñ 2.71 (5) 1.30 ñ 0.17 (3) -- Male 3.74 +- 0.77 (25) 2.62 +- 1.08 (13) 1.25 + 0.25 (2) Total 3.82 ñ 0.76 (30) 2.38 -+ 0.88 (16) 1.25 -+ 0.25 (2) All differencesamong ages within sexesand for totalswere not significant(Kruskal-Wallis or Mann-Whitneytests, P > 0.05).

= 110) were successfulversus only 4% (n = 48) The heightan individual'sfirst nest as a breed- of the nestshigher than 2.25 m aboveground ing adult was not significantlycorrelated with (X2 = 6.87, df = 1, P < 0.01;Fig. 1). Thus, suc- thatof its natalnest (Spearman correlation, rs = cessful nests were significantly lower than -0.144, n = 31, P = 0.77), and a negative cor- nests that failed (successfulnests, œ = 2.09 --- relation existed between an individual's mean SD of 3.15 m, n = 27; failed nests, œ = 3.63 -+ nestheight and the heightof its natalnest (r• = 3.90 m, n = 131; Mann-Whitney U-test, z = -0.359, n = 36, P < 0.05). Theseresults are in- 2.44, P < 0.02). Indeed, the highestof the two consistentwith the idea that the natal sitepro- successful nests that was more than 2.25 m vides cues for subsequentnest-site selection. aboveground (17 m; seeFig. 1) was attackedby Philopatricrecruits were both male (n = 30) a predatortwo daysbefore normal fledging age and female (n = 6), but no significantrelation- of the young,but becausethree of thebrood of ship existedbetween heights of natal and later approximatelynine chicksescaped predation nestswhen eachsex was analysedseparately and fledged,this nestwas classified as success- (Spearmancorrelations, all P > 0.05). ful. If individuals learn the relative quality of Nests in forks were less successful than those nestsites, a progressivechoosing of lowernests in peripheralbranches (forks, 0% success,n = might be predicted as birds became older 22; branches,20% success,n = 136; X2 = 3.96, However,nest height did not decreasesignifi- df = 1, P < 0.05).Nests in protectivevegetation cantly as the number of nestsbuilt by an indi- tended to be more successful than those in un- vidual increased(nest height vs. nest order; protected vegetation,although the difference males, r, = 0.027, n = 64, P = 0.83; females, rs was not significant(protective, 21% success,n = 0.136, n = 17, P = 0.65). Furthermore,for in- = 110; unprotected, 8% success,n = 48; X2 = dividuals of known age (i.e. birds banded as 2.90, df = 1, P < 0.09). Nest height and plant nestlingswho subsequentlybred in the study speciesclearly are not independent,but height area) there was no significanteffect of age on was probablythe moreimportant factor; none nestheight, using either the meanheight of all of the nestsin protectivevegetation above 2.25 nestsin eachyear, or just the first nestbuilt in m was successful(n = 10), whereas 23% of eachyear (Table1). The apparentabsence of an thosebelow 2.25 m succeeded(n = 100;Fisher's ageeffect on nest height should be treatedwith exact test, P = 0.08). caution,however, because a nonsignificantten- Factorsinfluencing nest-site selection.--The na- dencyexisted for nest height to declinewith tal nest may be a reliable cue to first-time age (Table1). The samplesize for femaleswas breedersas to what constitutesa goodnest site. very small, and for males and both sexescom- Weknew the firstnest site of 31 fledglingsfrom bined,the power of the testsat detectinga "me- the studyarea that recruitedinto thebreeding dium" effect of age on mean nest height be- populationin a subsequentyear. The laternests tweenthe first and secondyear was only 0.42 of a further five recruits also were recorded. and 0.48,respectively (Cohen 1988). April 1999] Experienceand Nest-site Selection 359

TABLE2. Seasonalchange in the numberof Long-tailedTit nestsbuilt aboveor below2.25 m and in the heightof nests(œ _+ SE). The frequencydistribution of the numberof nestsbuilt in eachheight category variedsignificantly with nestorder (X 2 = 21.6,df = 2, P < 0.001).

Nest order Heightcategory First Second Third Below 2.25 m 89 19 11 Above 2.25 m 21 25 7 Meanheight (m) 2.58 + 0.30 5.03 + 0.66 3.69 _+0.99

Nestheight changed dramatically with sea- males,X 2 = 0.32, df = 1, P = 0.57; pairs, X2 = son(Table 2) and was not associatedwith bud- 0.65,df = 1, P = 0.42),i.e. failureat oneheight burst or other seasonaldifferences in vegeta- did not result in a switch to the other height tion.Previous studies have suggested that such category. changesare a responseto theexperience of fail- BecauseLong-tailed Tits are singlebrooded, ure or successof previousnests (Lack and Lack we could not compareinternest distancesfol- 1958, Gaston1973, Glen 1985).A pair whose lowing a sequenceof successfuland unsuc- breeding attempt fails might change nest cessfulattempts in the sameyear. However, a height and/or moveto a new area for a sub- comparisonof the distancemoved by failed sequentattempt. For example,Long-tailed Tits breedersfollowing nest predation(œ = 175 q- may placetheir first nestof the year in what is 132 m, n = 49) or abandonment (œ= 230 q- 249 perceivedto be a "good" location(i.e. a low m, n = 15) showedthat birds did not respond site)but respondto nestfailure by switchingto differently to these two causesof failure a differentlocation (i.e. a high site)for subse- (Mann-WhitneyU-test, z = 0.357,P = 0.72). quentnests. In 64 cases,we recordedthe height Glen (1985)suggested that after a nestfailure, of a failednest and its replacementwithin the pairscould either move far awayfor theirnext sameseason. In 49 of thesecases, the pair re- attemptor stay in the samearea and switch mainedtogether for the secondattempt, but in nestsites, and he predicteda negativerelation- the remainderthe pair divorcedand the re- shipbetween the changein nestheight and in- placementnest of one or both membersof the ternestdistance. In our study,no suchcorrela- original pair was recorded.No significantas- tionexisted for all nestchanges (r• = 0.169,n = sociationof height categorieswas evident,ei- 49, P = 0.24) nor for nest changesfollowing ther positiveor negative,between successive predation(r• = 0.282,n = 37, P = 0.09).There- nests (males, X2 = 0.54, df = 1, P = 0.46; fe- fore,the variation in nestheight within seasons remainsunexplained. Breedingexperience had no effecton nest- TABLE3. Effect of successfulreproduction on nest heightin maleand female Long-tailed Tits. Values siteselection across seasons. The meanheights are nestheight (:f _+SE, with n in parentheses)be- of nests built by individuals before and after fore and after an individual's first successful nest their first known successfulbreeding attempt for:(1) all birdsthat bred successfully atleast once, were compared in a population analysis in- and(2) a pairedcomparison of birdsthat bred suc- cludingall successfulbreeders and in a paired cessfullyat leastonce and for which nestheights were recorded before and after the successful at- comparisonof meannest heights for the same tempt. No differences in before and after nest individuals before and after their first success- heightswere significant(Mann-Whitney U-test ful breedingattempt (Table 3). In eachcompar- and Wilcoxonsigned-rank test). ison,experience had no significanteffect for ei-

Sex Before After thersex. These results suggest that Long-tailed Tits do not respondto the experienceof suc- All successful breeders cessfulreproduction in low nestsby building Male 2.34 ñ 0.42 (24) 2.31 + 0.64 (13) Female subsequentnests lower down than before that 1.91 + 0.32 (27) 1.65 + 0.40 (12) SUCCESS. Paired comparison To testfor an effectof helpingexperience on Male 2.45 ñ 0.69 (13) 2.31 • 0.64 (13) Female nest-siteselection, the mean height of nests 2.44 + 0.68 (12) 1.65 -+ 0.40 (12) built by malesbefore and after their first ex- 360 HATCHWELLET AL. [Auk, Vol. 116 perienceof helpingat a successfulnest was ex- lected lower sites for their own subsequent amined. As before, two analyses were per- nests. formed, a population analysis including all The nesting successrate of 17% was very helpersat successfulnests, and a paired com- similar to resultsof previousstudies of Long- parisonof meannest heights for the samein- tailed Tits conductedin similar habitats (Lack dividualsbefore and after their first experience and Lack 1958, Riehm 1970, Gaston 1973, Glen of helping.The population comparison showed 1985).Small open-nestingpasserines often suf- that the meannest height after helping(1.61 q- fer high ratesof nest predationin woodlands 1.05m, n = 9) wassignificantly lower than be- (Ricklefs 1969, O'Connor and Shrubb 1986, fore helping (3.82 q- 3.11 m, n = 18; Mann- Moller 1988,Hatchwell et al. 1996),but the pro- Whitney U-test,z = 2.72,P < 0.01).Similarly, portion of nests lost to predators by Long- in the paired comparisonof the sameindivid- tailed Tits was particularly high. Nest-site uals, nest height after helping at a successful characteristicshad a significanteffect on the nest(• = 1.90 _+1.02 m, n = 7) wassignificantly successof Long-tailed Tit breeding attempts. lowerthan that before helping (• = 5.59+- 3.75 Nestheight was the most important factor mea- m, n = 7; Wilcoxonsigned-rank test, z = 2.37, sured, and the vegetationproviding support P < 0.02). These resultsindicate that helpers for nestshad little additionaleffect on nesting learn what constitutesa goodnest site during success.The effectof nestheight on reproduc- theirperiod of helping.Too few femalehelpers tive successwas consistentwith previousstud- were presentto testfor an effectof helpingex- ies, which have shown similar height profiles periencein females. and a similardecline in successwith increasing Somemales whose breeding attempt failed nestheight (Lackand Lack 1958,Riehm 1970, did not help at the nestof anotherpair, and in- Gaston1973). This consistent effect of heighton stead joined a flock of failed breeders.The the successof breedingattempts raises an ob- heightof nestsof these"nonhelpers" in sub- viousquestion: why do Long-tailedTits contin- sequentseasons (• = 3.41 q- 2.19 m, n = 10) did ue to build nestsin predictablyunsuccessful not differ significantlyfrom that prior to their sites given the strong selection pressure failed breeding attempt (• = 2.93 +- 3.96 m, n against high nests?Several explanations are = 10; Wilcoxonsigned-rank test, z = 0.95,P = possiblefor this apparently maladaptivebe- 0.34). Theseresults support the previouscon- havior:(1) suitablesites are in shortsupply; (2) clusionthat the experiencegained by helping nest building in high sites is maintainedby during a successfulbreeding attempt influenc- geneflow from areaswhere high nestsare rel- es subsequentchoice of nestsites. Furthermore, atively successful;(3) high nest sites may be more successfulin someyears than in others; the experienceof helpinghad an effecton sub- and (4) the ability of breedersto choosea suc- sequentreproductive success. Three of seven cessfulsite may be limited if theyhave little in- (43%) failed breederswho helped were suc- formationas to what constitutesa good site. cessfulin the subsequentseason, whereas none Suitablesites in shortsupply.--Long-tailed Tits of the 10 failed breederswho did not help was are relatively unconstrainedin their choiceof successfulin the followingseason (Fisher's ex- nest site becausethey occur in low densities act test, P = 0.05). and are not territorialduring the breeding sea- son, exceptfor a small area immediatelysur- DISCUSSION rounding the nest (Cramp and Perrins 1993). Defenseof thenesting area by establishedpairs In the Long-tailedTits we studied,the main declinesas the seasonprogresses, and in three causeof breeding failure was predation,and instanceslate replacementnests were built the probability of predation was higher for within 20 m of active nests. Because a shrub nests above 2.25 m than for nests below this layer of holly and brambles was extensive height.The choiceof nestheight by an individ- throughoutmuch of the study site, low nest ual wasnot influencedby its age at the time of siteswere abundant,and birds did not appear the attempt,by the heightof its natalnest, nor to be forcedto attemptbreeding in unsuccess- by its experienceas a successfulbreeder. How- ful sites(e.g. conifers or tree forks)in any part ever,males who helpedat successfulnests se- of the study area. Our contentionthat suitable April 1999] Experienceand Nest-site Selection 361

low nestsites were available to all pairsis fur- creaseas birds grew older.Thus, some effect of thersupported by thefact that pairs occupied experiencemay be weakly correlated with age. largebreeding home ranges (f = 4.37 _+1.91 Third,an individual'sexperience of success- ha,n = 17)and often moved long distances be- ful reproductionmight influencenest-site se- tweensuccessive attempts within a season(œ = lectionthrough a mechanismof trial anderror, 188 __166 m, n = 64, range24 to 1,000m). asshown in thePinyon (Gymnorhinus cyan- Geneflow from other habitats and temporal vari- ocephalus;Marzluff 1988).No supportexisted ationin success.--Althoughhigh nests were rel- forthe hypothesis that an individual's personal ativelyunsuccessful in thisstudy, it is possible reproductiveexperience influenced its nest-site thatin adjacentareas or in someyears high selection(Table 3). However,the cooperative nests do relativelybetter Either possibility behaviorof Long-tailedTits offersa further couldmaintain apparently realadaptive selec- sourceof experience;viz. the "skills" hypoth- tionof nestsites in ourstudy population. Long- esis of Brown (1987).In typical cooperative tailedTits aresedentary in nature(Cramp and breeders,helping precedes dispersal and in- Perrins1993), and philopatricrecruitment is dependentbreeding (Emlen 1991), so helpers highwithin our study area (about 20% of fledg- maygain direct fitness by improvingtheir par- lingsrecruit into the study population). The av- entingskills, e.g. in nestbuilding, incubation, erage distancebetween an individual's natal or theprovisioning of nestlings.In oneof the nestand its firstbreeding attempt was 522 m few studiesto demonstrateskill acquisition for femalesand 693 m for males(B. J. Hatchwell throughhelping, Komdeur (1996) showed that unpubl.data). This low dispersalsuggests that femaleSeychelles Warblers with helpingexpe- extensivegene flow into the studypopulation rience constructed better nests than females of from areas with radically different habitat thesame cohort but with no priorexperience. structureis unlikely.Moreover, previous stud- Nestsbuilt by birds with helpingexperience ies conducted over several seasons have found were lesslikely to fall out of trees,and their re- similarsite effects to thosereported here (Lack productivesucces was consequentlyhigher and Lack1958, Riehm 1970, Gaston 1973), and Our resultssupport the hypothesisthat the ex- in thisstudy we foundno significantannual periencegained from helping has a positivein- variationin successrates of nestswith respect fluenceon nest-siteselection. Long-tailed Tits to nestheight. Therefore, we considerit unlike- are atypicalcooperative breeders because they ly thateither of thepossibilities noted above ex- mayhave repeated opportunities for helping in plainsthe choice of unsuccessful sites by Long- successiveyears. Therefore, any tendencyfor tailedTits in our studypopulation. nest heightto decreasewith age (Table1) Learningand a directfitness benefit from help- would be consistentwith experiencein nest ing.--Finally,we considerthe question of howin- placementgained through helping. This benefit dividualsmay gain 'information about the relative of helpingbehavior was reflected in a higher qualityof differenttypes of nestsite. In theap- probabilityof successfulreproduction in sub- parent absenceof an innatepreference for the sequentseasons, although the samplesize was most successfulsites, what cues are available to small,and it is possiblethat individual quality birdswhen deciding where to buildtheir nests? is a confoundingfactor that influencessuccess. One possibilityis that the natal nest is used Long-tailedTit helpersare knownto accruein- asa templatefor subsequentchoices as a breed- directfitness benefits through their cooperative er. However, no correlation existed between behavior(Hatchwell and Russell 1996), but the heightsof natal nestsand first nests,nor be- effectof helpingexperience on nest-siteselec- tweennatal nestsand the meanheight of all tion may representa significantsource of direct subsequentnests. Second, an age-related fitnessbenefits for helpersin thisspecies. changemay occurin nest-siteselection if, for Why should the experienceof success example, individuals acquire information throughhelping at the nest of a conspecific,but aboutpredator behavior or aboutwhat consti- not the personalexperience of an individual's tutesa goodor bad nestsite throughexperi- own success, influence nest-site selection? The enceor observationof neighboringnests. No most likely explanationis that althoughlow supportwas apparentfor this hypothesis,al- nestsare more successful than high nests, they thougha tendencyexisted for nest height to de- stillhave a veryhigh failure rate (Fig. 1). Thus, 362 HATCHWELLET AL. [Auk, Vol. 116 successfulbreeders may haveexperienced suc- BAEYENS,G. DEN.1981. Magpie breedingsuccess and cesswith a low nest,but a high probabilityex- Carrion Crow interference. Ardea 69:125-140. ists that they also will have experiencedthe BOULINIER, t., E. DANCHIN, J.-Y. MONNAT, C. DOU- failure of low nests. In fact the ratio of failures: TRELANT,AND g. CADIOU.1996. Timing of pros- successeswas 3:1 even for low nests, so this cri- pectingand the valueof informationin a colo- nial breeding bird. Journalof Avian Biology27: terion would providea ratherpoor cue to the 252-256. attributesof a goodnest site. BROWN,J. L. 1987.Helping and communalbreeding Experiencethrough helping providesa less in birds. PrincetonUniversity Press,Princeton, ambiguouscue. Helpers arrive at neststo help New Jersey. only duringthe nestling period. Nesting failure CHAMBERLAIN, D. E., B. J. HATCHWELL, AND C. M. at this stageis muchlower (44%), and an even PERRINS.1995. Spacedout nestsand predators: smaller proportionof nestsfail after helpers An experimentto testthe effects of habitatstruc- have arrived (18%). Therefore,learning based ture. Journalof Avian Biology26:346-349. onhelping experience offers a morereliable cue CLUTTON-BROCK,t. H. (Ed.). 1988. Reproductive aboutwhat constitutesa good nest site.This success: Studies of individual variation in con- cue is available mainly to males becausefe- trastingbreeding systems. University of Chica- go Press,Chicago. males rarely help (Glen 1985,B. J. Hatchwell COHEN,J. 1988.Statistical power analysis for the be- unpubl. data). In this regard, it is interesting havioral sciences. Lawrence Erlbaum Associ- that Riehm (1970) noted from behavioral ob- ates,Hillsdale, New Jersey. servationsthat malesassume the primary role COLLIAS,N. E., AND E. C. COLLIAS. 1984. Nest build- in nest-site selection. ing and bird behavior.Princeton University In conclusion,our resultsindicate that Long- Press,Princeton, New Jersey. tailedTits learn aboutnest-site quality through COLWELL,M. A. 1992.Wilson's Phalarope nest suc- the experienceof helping at successfulnests cess is not influencedby vegetationconceal- and not througha processof trial and errorus- ment. Condor 94:767-772. ing their own nests.Nevertheless, no reasonis CRAMP,S., AND C. M. PERRINS(Eds.). 1993. The birds obviouswhy the choiceof nestsites with very of the western Palearctic, vol. 7. Oxford Univer- low probabilityof successpersists despite the sity Press,Oxford. strongselection against such apparently mal- CRESSWELL,W. 1997a. Nest predation: The relative ef- fectsof nest characteristics,clutch size and pa- adaptivebehavior. Perhaps the most likely ex- rental behaviour. Animal Behaviour 53:93-103. planationis that althoughlow nestsare more CRESSWELL,W. 1997b.Nest predationrates and nest successfulthan high nests,the majorityof low detectability in different stagesof breeding in nestsstill fail. Therefore,experience of frequent Blackbirds Turdusmerula. Journal of Avian Biol- reproductivefailure in low nestsmay result in ogy 28:296-302. selectionof alternativenest sites, even though CURIO,E. 1983.Why do youngbirds reproduceless their success rate is even lower. well? Ibis 125:400-404. DESROCHERS,g. 1992a.Age and foraging successin ACKNOWLEDGMENTS European Blackbirds: Variation between and within individuals. Animal Behaviour 43:885- We thank Tim Birkhead, Glen Woolfenden, Jan Ek- 894. man, and two anonymousreferees for their valuable DESROCHERS,A. 1992b. Age-related differences in re- commentson themanuscript; and Hallam GolfClub, production by European Blackbirds:Restraint YorkshireWater, and SheffieldCity Councilfor per- or constraint?Ecology 73:1128-1131. mission to watch birds on their land. This work was EMLEN,S. t. 1991.Evolution of cooperativebreeding fundedby grantsfrom the Associationfor the Study in birds and mammals.Pages 301-335 in Behav- of Animal Behaviour, Nuffield Foundation, Univer- ioural ecology:An evolutionaryapproach, 3rd sity of Sheffield,and the Natural EnvironmentRe- ed. (J.R. Krebsand N. B. Davies,Eds.). Blackwell searchCouncil, for which we are mostgrateful. Scientific Publications, Oxford. FILLIATER, t. S., R. BREITSWISCH,AND P. M. NEALEN. LITERATURE CITED 1994. Predation on Northern Cardinal nests: Does choice of nest site matter? Condor 96:761- ANDERSSON,M., AND G. WIKLUND.1978. Clumping 768. versus spacingout: Experimentson nest preda- GISTON,A. J.1973. The ecologyand behaviour of the ,ion in Fieldfares(Turdus pilaris). Animal Behav- Long-tailedTit. Ibis 115:330-351. iour 26:1207-1212. GLEN,N. W. 1985.The cooperativebreeding behav- April 1999] Experienceand Nest-site Selection 363

iour of the Long-tailedTit (Aegithaloscaudatus). lation to nest sites, nest predation, and food. Ph.D.dissertation, University of Oxford,Oxford. EcologicalMonographs 65:101-127. GLEN,N. W., ANDC. M. PERRINS.1988. Cooperative MARZLUFF,J. M. 1988. Do Pinyon Jays alter nest breeding by Long-tailed Tits. British Birds 81: placementbased on prior experience?Animal 630-641. Behaviour 36:1-10. HANSELL,M. 1993. The Long-tailed Tit's nest of MARZLUFF,J. M., ANDR. P.BALDA. 1992. The Pinyon many parts. BTO News 186:20. Jay:Behavioral ecology of a colonialand coop- HATCHWELL,B. J. 1991.An experimentalstudy of the erative corvid. T & A D Poyser,London. effectsof timingof breedingon thereproductive MAYER-GRoss, H., H. Q. P. CRICK, AND J. D. GREEN- successof Common Guillemots (Uria aalge). wooo. 1997.The effectof observersviEting the Journalof Animal Ecology60:721-736. nestsof :An experimentalstudy. Bird HATCHWELL,B. J., D. E. CHAMBERLAIN,AND C. M. Study 44:53-65. PERRINS.1996. The reproductive successof MAYFIELD,H. 1975. Suggestionsfor calculatingnest Blackbirds Turdus merula in relation to habitat success. Wilson Bulletin 87:456-466. structure and choice of nest site. Ibis 138:256- MEILVANG, D., A. MOSKNES, AND E. ROSKAFT. 1997. 262. Nest predation,nesting characteristics and nest HATCHWELL,B. J., AND A. E RUSSELL.1996. Provi- defencebehaviour of Fieldfaresand Redwings. sioning rules in cooperativelybreeding Long- Journalof Avian Biology28:331-337. tailed Tits Aegithaloscaudatus: An experimental MOLLER,A. P. 1988. Nest predation and nest site study.Proceedings of the RoyalSociety of Lon- choicein passerinebirds in habitatpatches of don Series B 263:83-88. different size: A study of and black- HEINSOHN, R. G., A. COCKBURN,AND R. B. CUNNING- birds. Oikos 53:215-221. HAM. 1988. Foraging, delayed maturation, and O'CONNOR,R. J., ANDM. SHRUBB.1986. Farming and advantagesof cooperativebreeding in White- birds. CambridgeUniversity Press,Cambridge, winged Corcorax melanorhamphos. United Kingdom. Ethology77:177-186. PICMAN, J., M. L. MILKS, AND M. LEPTICH.1993. Pat- HOLWAY, D. A. 1991. Nest site selection and the im- ternsof predationon passerinenests in marshes: portance of nest concealmentin the Black- Effectsof water depth and distancefrom edge. throated Blue Warbler. Condor 93:575-581. Auk 110:89-94. KOMDEUR,J. 1996.Influence of helpingand breeding POTTS,G. R., J. C. COULSON,AND I. R. DEANS. 1980. experienceon reproductiveperformance in the Populationdynamics and breeding successof SeychellesWarbler: A translocationexperiment. the Shag, Phalacrocoraxaristotelis, on the Farne BehavioralEcology 7:326-333. Islands, Northumberland. Journal of Animal LACK, D. 1954. The natural regulation of animal Ecology49:465-484. numbers. Clarendon Press, Oxford. RICKLEFS,R. E. 1969.An analysisof nestingmortality LACK,D., AND E. LACK. 1958. The nesting of the in birds. SmithsonianContributions to Zoology Long-tailedTit. Bird Study 5:1-19. 9:1-48. LENINGTON, S. 1979. Predators and blackbirds: The RIEHM, H. 1970. Okologie und Verhalten der "Uncertainty Principle"in field biology.Auk 96: Schwanzmeise(Aegithalos caudatus L.). Zoologis- 190-192. cheJahrbuecher Systematick 97:338-400. MARTIN,t. E. 1995.Avian life historyevolution in re- AssociateEditor: J. Ekman